{"day1":[{"topic":"主题演讲","content":[{"id":"1070","conid":"18","topic":"主题演讲","title":"普适智能和普适学习：智能革命和智能经济的引擎","introduce":"<p>Pervasive Intelligence and Pervasive Learning: The Engine of Intelligent Revolution and Intelligent Economy<\/p>\n\n<p>本报告从技术和历史角度指出虽然人工智能和机器学习取得了很多成就，但总体还处于智能革命的前夜。智能革命将比农业革命和工业革命对人类的影响更深远，新的智能经济模式也将产生。2-5年后可能新的一波人工智能和机器学习技术将兴起，突出表现为：<\/p>\n\n<p>1）机器学习从云端走向各类本地智能终端\/传感器\/设备，云端机器学习（比如深度学习等方法）和本地机器自主学习（比如超限学习机，Extreme Learning Machines (ELM) 等方法）有机融合；<\/p>\n\n<p>2）机器学习算法本身不再依赖于GPU的支撑，但GPU等却可以实现众多智能系统在云端的同步协同；<\/p>\n\n<p>3）机器学习和生物学习有逐步汇合的趋势；<\/p>\n\n<p>4）基于神经形态、FPGA和光技术的芯片开始在普适学习\/普适智能中显现魅力；<\/p>\n\n<p>5）冯&middot;诺依曼计算机体系结构不再是智能实现的必然；<\/p>\n\n<p>6）数据虽然不可缺少，但智能不必依赖于大数据，数据大也会造成机器学习过拟合。普适学习和普适智能是智能革命和智能经济的引擎。本报告也将分析人工智能的10大应用前景和智能革命的10大影响。<\/p>\n","guestName":"黄广斌","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/59527e58a1983.png","guestIntroduce":"0","beginTime":"1499391000","endTime":"1499393700","address":"大宴会厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1497525118","updatedAt":"1497525118","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"新加坡南洋理工大学","guestPosition":"教授","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/919\/595f21bd2cb80.pdf\/18"},{"id":"1062","conid":"18","topic":"主题演讲","title":"Data Outsourcing in Cloud Computing: Reliability, Security and Privacy","introduce":"<p>The many advantages of cloud computing architecture are increasingly attracting individuals and organizations to outsource their data from local to remote cloud servers. It is a clear trend that cloud data outsourcing is becoming a pervasive service. Along with the widespread enthusiasm on cloud computing, however, concerns on data reliability, security and privacy with cloud data storage are arising. For protecting data privacy, sensitive data has to be encrypted before outsourcing, which obsoletes traditional data utilization based on plaintext keyword search. Thus, enabling an encrypted cloud data search service is of paramount importance.<\/p>\n\n<p>This talk will first give an overview on data outsourcing in cloud computing, and then discuss challenges and corresponding solutions of reliability, security and privacy which raise as primary obstacles to the adoption of cloud computing. At last, it will present latest research on generic computation outsourcing in cloud computing.<\/p>\n\n<h2>Agenda<\/h2>\n\n<ul>\n\t<li>Introduce data outsourcing in cloud computing<\/li>\n\t<li>Outsource data in cloud computing\n\t<p style=\"margin-left:2px\">a.Security and reliability issues<\/p>\n\n\t<p style=\"margin-left:2px\">b.Reliable storage: network coding vs fountain codes<\/p>\n\n\t<p style=\"margin-left:2px\">c.Bilinear map based public integrity verification<\/p>\n\t<\/li>\n\t<li>Search over encrypted data in cloud computing\n\t<p style=\"margin-left:2px\">a.Privacy issues (data, index, search)<\/p>\n\n\t<p style=\"margin-left:2px\">b.Privacy-preserving searchable encryption<\/p>\n\n\t<ol style=\"list-style-type:lower-roman\">\n\t\t<li>Basic symmetric\/asymmetric cipher<\/li>\n\t\t<li>Order preserving encryption<\/li>\n\t\t<li>Predicative encryption<\/li>\n\t<\/ol>\n\t<\/li>\n\t<li>Generic computation outsourcing in cloud computing<\/li>\n<\/ul>\n\n<h2>For audience<\/h2>\n\n<ul>\n\t<li>Overview reliability, security and privacy issues when adopting cloud computing architecture<\/li>\n\t<li>Know latest cutting-edge research going on in these fields<\/li>\n<\/ul>\n","guestName":"曹宁","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/59253cc381343.png","guestIntroduce":"0","beginTime":"1499394600","endTime":"1499397300","address":"大宴会厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1496906982","updatedAt":"1496906982","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"WalmartLabs","guestPosition":"Engineering Manager","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/918\/595f1be103c83.pdf\/18"},{"id":"1096","conid":"18","topic":"主题演讲","title":"Cloud vNext：正在兴起的云计算新技术","introduce":"<p>Cloud vNext：The Emerging New Cloud Technologies&nbsp;<\/p>\n\n<p>云计算的兴起影响了许多行业，经过这十来年的发展，我们看到IT行业的格局已经完全被云计算颠覆，同样像CT这样的行业也正在经历云计算技术所带来的巨大改变和机遇。与此同时，随着云计算应用的深入，随着人工智能、物联网等应用的逐步推广，云计算平台自身的架构和技术也在逐渐发生着变化，以应对新的需求和技术发展。<\/p>\n\n<p>本报告将从云计算发展变化开始阐述云计算发展的几个大的趋势，然后结合人工智能、物联网等应用场景的特点阐述其对底层云平台的技术需求，并结合近期计算、存储和网络等软硬件的发展，分析云计算下一步的几个技术发展大趋势。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>云计算行业的发展概况<\/li>\n\t<li>新应用场景对云的需求以及技术应对<\/li>\n\t<li>新的软硬件技术发展及云的技术发展趋势<\/li>\n<\/ul>\n","guestName":"方国伟","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5953ac8b7e9c8.png","guestIntroduce":"0","beginTime":"1499398200","endTime":"1499400900","address":"宴会厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1498817953","updatedAt":"1498817953","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"华为","guestPosition":"中央软件院 架设部部长","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/917\/595f155a0fd87.pdf\/18"}],"producter_name":"0","topic_id":"102","order":"0","topic_description":"","introduce":"","guestavatar":"","company":"","position":""},{"topic":"实时处理技术与创新业务架构（厂商共建）","content":[{"id":"1082","conid":"18","topic":"实时处理技术与创新业务架构（厂商共建）","title":"腾讯信鸽实时精准推送系统的演进与实践","introduce":"<p>移动消息推送是触达用户的重要方式。信鸽作为免费的推送工具，服务了腾讯及腾讯外很多头部应用。如何维持亿级长连接，海量、实时、精准推送，有不小挑战。本次重点分享如何挖掘单机潜力，构建分布式消息推送系统。<\/p>\n\n<h2>听众收益<\/h2>\n\n<ul>\n\t<li>了解如何对单机性能进行挖掘，构建海量长连接接入服务<\/li>\n\t<li>了解实时精准推送系统的构建<\/li>\n<\/ul>\n","guestName":"甘恒通","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5951fea330102.png","guestIntroduce":"0","beginTime":"1499391000","endTime":"1499393700","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1498545924","updatedAt":"1498545924","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"腾讯","guestPosition":"数据平台部后台开发高级工程师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/914\/595ec5eaaa077.pdf\/18"},{"id":"1086","conid":"18","topic":"实时处理技术与创新业务架构（厂商共建）","title":"质量实时监控：全球音视频实时传输的关键帧","introduce":"<p>本次演讲主要分享如何通过大数据平台解决全网音视频传输的实时质量监控问题。<\/p>\n\n<p>在全球实时音视频传输过程中，会面临很多实时监控挑战，比如设备、服务器全球分布很广；移动设备种类太多；数据指标不够详尽；时间序列的存储需要优化等等。<\/p>\n\n<p>声网Agora.io的全球虚拟实时通信网SD-RTN&trade;支撑全球200多个国家的实时通话和直播，从Day One开始，团队希望以数据为核心来驱动质量改进和优化，因此在数据架构设计、数据收集、分析、还原、预警和使用上有很多实践经验分享，同时总结出如何将数据与业务紧密结合的&ldquo;关键帧&rdquo;。&nbsp;&nbsp;<\/p>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>能够了解基于&ldquo;数据驱动业务&rdquo;的大数据平台-实时监控平台的架构设计原理&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;<\/li>\n\t<li>能够了解大数据平台-实时监控平台在数据架构设计、收集、分析、还原、预警等多方面的实践案例分享 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;<\/li>\n\t<li>能够了解如何真正将数据与业务紧密结合，驱动业务发展的难点和优化方案。<\/li>\n<\/ul>\n","guestName":"何丰","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/594bdb99375bb.png","guestIntroduce":"0","beginTime":"1499394600","endTime":"1499397300","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1498549135","updatedAt":"1498549135","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"声网Agora.io","guestPosition":"高级架构师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/915\/595ef974b51c2.pdf\/18"},{"id":"1093","conid":"18","topic":"实时处理技术与创新业务架构（厂商共建）","title":"大疆服务网关全球化设计和实践","introduce":"<p>随着业务的扩展，每个公司都必须依托自身的业务情况推动系统架构的演进与优化。在微服务当道的年代，一家从事无人机全球业务的公司，如何构建互联网服务化之路？又是如何出海，在全球范围内搭建服务网关的呢？<\/p>\n\n<p>此次分享主要介绍：从2015年建立至今，大疆是如何打造一个接地气、符合自身业务特性的网关，以及如何设计覆盖全球范围的服务治理、服务监控，从而保证服务的性能、健壮和稳定的。比如：网关权限管理的颗粒度如何规划？多地域之间的服务路由要如何设计？如何保证接入服务的安全？<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>开天辟地：单体应用的春天<\/li>\n\t<li>万物生长：多语言、多服务<\/li>\n\t<li>千头万绪：API请求的黑匣子<\/li>\n\t<li>整顿干坤：服务网关的光环<\/li>\n<\/ul>\n\n<h2>听众收益<\/h2>\n\n<ul>\n\t<li>在实现微服务架构初期关键的决策因素<\/li>\n\t<li>全球多地域网关系统设计借鉴和参考<\/li>\n<\/ul>\n","guestName":"张清平","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5954ba2a926c9.png","guestIntroduce":"0","beginTime":"1499398200","endTime":"1499400900","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1498725613","updatedAt":"1498725613","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"DJI大疆创新","guestPosition":"架构师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/984\/5965f83d13738.pdf\/18"}],"producter_name":"薛梁","topic_id":"192","order":"3","topic_description":"<p>伴随着市场的成熟，如何提供更好的服务、更快的数据决策，成为竞争的关键点。实时计算技术作为其中的一项关键技术，开始在业界广泛流行。实时处理要求系统有很高的可靠性，低延迟、高时效，信息的任何错误或丢失都可能造成严重的损失。挑战常在，怎么应战？另外，在硬件、终端软件和后端服务联动的业务架构中，有哪些创新的架构设计及搭建方法？<\/p>\n\n<p>因此，本专题将邀请来自不同公司技术团队的专家和领导，分享他们在高并发与实时处理系统架构设计与优化过程中的经验与心得，讨论方向将聚焦于：如何处理实时计算中遇到的各种问题，保证数据的效率和正确？如何通过大数据平台解决全网音视频传输的实时质量监控问题？以及业务架构变革之道等。<\/p>\n","introduce":"伴随着市场的成熟，如何提供更好的服务、更快的数据决策，成为竞争的关键点。实时计算技术作为其中的一项关键技术，开始在业界广泛流行。实时处理要求系统有很高的可靠性，低延迟、高时效，信息的任何错误或丢失都可能造成严重的损失。挑战常在，怎么应战？另外，在硬件、终端软件和后端服务联动的业务架构中，有哪些创新的架构设计及搭建方法？","guestavatar":"http:\/\/static.geekbang.org\/ck\/59520e41b5c0e.png","company":"InfoQ","position":"策划编辑"},{"topic":"高效运维与微服务（厂商共建）","content":[{"id":"1085","conid":"18","topic":"高效运维与微服务（厂商共建）","title":"运行在Amazon ECS上的微服务架构和实践","introduce":"<p>运行和管理大型可伸缩的微服务化架构的集群一般会面临各种各样的挑战，容器化常常是这个挑战的很重要的选择。<\/p>\n\n<p>本次将会为大家介绍Amazon EC2 Container Service (ECS)如何在高可用，高扩展方面能更好地帮助我们的客户运行和维护Docker的容器；<\/p>\n\n<p>并且我们将会探讨为何我们需要微服务架构，以及微服务架构在ECS上运行的一些实践；<\/p>\n\n<p>最后将会分享来自互联网的客户使用ECS服务搭建内部的微服务架构的平台中的种种最佳实践，以及来自客户在蓝绿部署和微服务对数据库管理上与ECS结合的案例分享。<\/p>\n","guestName":"张孝峰","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/596597494de62.png","guestIntroduce":"0","beginTime":"1499391000","endTime":"1499393700","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0"},"createdAt":"1498549011","updatedAt":"1498549011","status":"0","days":"07","spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0","guestCompany":"AWS","guestPosition":"解决方案架构师","dayorder":1,"ppt_link":""},{"id":"1084","conid":"18","topic":"高效运维与微服务（厂商共建）","title":"无惧挑战： 一站式构建轻运维高可用架构","introduce":"<p>业务高可用和数据高可靠的云产品如何搭建？如何将用户的运维成本降至趋近于零？本次演讲将深入剖析通过多层架构和服务抽象的方式，提供从接入到网络到存储的一站式高可用体系。从物理环境和虚拟环境带来的复杂挑战，到落地的技术应对和技术创新，UCloud将分享大量用户运营积累之后的扎实经验，点明数据高可靠和持续保护上的技术难点，掘出最优实现路径。<\/p>\n\n<p>&nbsp;<\/p>\n\n<h2>听众收益<\/h2>\n\n<ul>\n\t<li>复杂网络环境下高可用的最佳实践和经验分享&nbsp;<\/li>\n\t<li>云上用户数据不间断保护的意义、难度、价值和技术实现<\/li>\n\t<li>UCloud高可用架构的层次、架构、能力和边界&nbsp;&nbsp;<\/li>\n<\/ul>\n","guestName":"朱锋","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/594bdc4ba2f0e.png","guestIntroduce":"0","beginTime":"1499394600","endTime":"1499397300","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1498548688","updatedAt":"1498548688","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"UCloud","guestPosition":"高级产品经理","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/916\/595efb245b6a8.pdf\/18"}],"producter_name":"曹倩芸","topic_id":"191","order":"1","topic_description":"<p>随着运维受到越来越多的重视，运维体系也逐步丰富，出现了 DevOps 等理念将研发、测试、运维等流程连接起来。而容器技术更是从底层重构了运维，连接了开发、测试、部署、运行和监控全流程，进一步推动了运维体系从工具化逐步往平台化、自动化和智能化方向迁移。另一方面，微服务配置管理、容器化部署、自动化测试、微服务治理、微服务监控、安全、故障容忍等领域也受到越来越多的关注。<\/p>\n\n<p>本专题将邀请来自不同公司技术团队的专家和领导，分享他们在高效运维、微服务化架构以及容器技术进化上的经验与心得，本方向的相关演讲将聚集于：越来越多相关工具的集成以及自动化持续测试如何能更有效地控制质量、保障安全、降低成本、控制风险、提升效率，更加高效的支持复杂的大型分布式应用？如何将分享业务高可用和数据高可靠的云产品搭建问题以及如何将用户的运维成本降至趋近于零？为何要微服务架构，以及微服务架构在ECS上运行的一些实践等。<\/p>\n","introduce":"随着运维受到越来越多的重视，运维体系也逐步丰富，出现了 DevOps 等理念将研发、测试、运维等流程连接起来。而容器技术更是从底层重构了运维，连接了开发、测试、部署、运行和监控全流程，进一步推动了运维体系从工具化逐步往平台化、自动化和智能化方向迁移。另一方面，微服务配置管理、容器化部署、自动化测试、微服务治理、微服务监控、安全、故障容忍等领域也受到越来越多的关注。\n","guestavatar":"http:\/\/static.geekbang.org\/ck\/5951f8f309054.png","company":"InfoQ","position":"策划编辑"},{"topic":"支付系统和金融架构演进（厂商共建）","content":[{"id":"1083","conid":"18","topic":"支付系统和金融架构演进（厂商共建）","title":"系统架构的加法与减法","introduce":"<p>付钱拉的系统经历了四次演变，在每一次的进化过程中，结合场景不断的优化的框架，从无到有，经历了把小系统做成大的过程，也经历了把大系统逐步精简细化，逐步平滑过度的阶段。其中经历的很多，也收获了很多。从架构设计，业务开发，技术运维多角度分析架构的扩展与精简。<\/p>\n\n<h2>听众收益<\/h2>\n\n<ul>\n\t<li>了解创业型公司的架构如何从无到有的过程，初期加法困难，如何基于现有架构，快速扩张，跟上业务发展，而后期又面临着快速扩张后带来的繁杂无序，架构又要经历减法，逐步梳理清晰脉络，跟随业务的增长。<\/li>\n\t<li>了解金融系统架构的业务特点及架构设计难点。 金融业务有其自身特点，安全，稳定是重中之中。如何在保证这些特点的基础上打造适合的金融框架。<\/li>\n\t<li>监控与运维，是与系统架构脱离不开的部分，直接关系到系统的健壮性和可维护性，架构中如何针对这些特点进行合理设计，做到高效运维与监控。<\/li>\n<\/ul>\n","guestName":"周恒","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/594bdc8dc3bd0.png","guestIntroduce":"0","beginTime":"1499405400","endTime":"1499408100","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1498546134","updatedAt":"1498546134","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"付钱拉","guestPosition":"高级架构师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/920\/595f226b8e814.pdf\/18"},{"id":"1087","conid":"18","topic":"支付系统和金融架构演进（厂商共建）","title":"支付体系架构与实践","introduce":"<p>在过去4年的时间里，作为面向亿级用户的大型时尚消费平台，美联集团历经着高速的业务增长和快速的业务演进。<\/p>\n\n<p>而其中最重要的基础业务平台，美联支付如何稳打稳扎、平滑演进，快速适应并高效支持着业务的复杂变化。<\/p>\n\n<p>我们从单一功能到完整体系、从臃肿单体Php演变为高性能高可靠可伸缩的分布式服务架构，于16年快速融合美丽说和淘世界支付体系，并在历年大促中保持无故障的出色表现，逐渐摸索出适应全集团复杂业务形态和变化的支付平台架构。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>支付系统演进介绍；<\/li>\n\t<li>支付系统架构实现；<\/li>\n\t<li>遇到问题和相关实践。<\/li>\n<\/ul>\n\n<h2>听众收益<\/h2>\n\n<ul>\n\t<li>了解支付系统的整体架构；<\/li>\n\t<li>了解支付系统核心服务实现和交互；<\/li>\n\t<li>支付系统架构演进过程中遇到的问题和经验。<\/li>\n<\/ul>\n","guestName":"陈宗（铁手）","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/594bdc1f56ede.png","guestIntroduce":"0","beginTime":"1499409000","endTime":"1499411700","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1498549259","updatedAt":"1498549259","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":" 美丽联合集团","guestPosition":"资深工程师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/925\/595f2e8444514.pdf\/18"},{"id":"1088","conid":"18","topic":"支付系统和金融架构演进（厂商共建）","title":"公有区块链与区块链即服务","introduce":"<p>区块链是目前一个比较热门的概念，从技术角度来看，这是一个牺牲一致性效率且保证最终一致性的的分布式的数据库，当然这是比较片面的,从经济学的角度来看，这种容错能力很强的点对点网络，恰恰满足了未来经济的一个潜在需求&mdash;&mdash;低成本的可信环境。<\/p>\n\n<p>区块链分公有区块链和联盟链，而公有区块链在 2016 年与联盟链分道扬镳，作为国内第一个推出公有区块链的公司，本次分享重点讲述公链的实践以及应用场合，讲述如何与传统网络服务相结合。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>区块链技术简要；<\/li>\n\t<li>公有区块链的核心要素和发展趋势；<\/li>\n\t<li>公有区块链技术实践&mdash;&mdash;Metaverse 元界；<\/li>\n\t<li>区块链即服务&mdash;&mdash;未来新架构模式 。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解区块链常见技术栈；<\/li>\n\t<li>了解目前区块链技术的局限性与发展趋势；<\/li>\n\t<li>了解公有区块链的架构设计与开源社区；<\/li>\n\t<li>了解如何使用公有区块链服务,发掘新的业务 。<\/li>\n<\/ul>\n","guestName":"陈浩","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/594bdbf8681f6.png","guestIntroduce":"0","beginTime":"1499412900","endTime":"1499408400","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1498549378","updatedAt":"1498549378","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"维优元界","guestPosition":"CTO","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/932\/595f3a87120b6.pdf\/18"},{"id":"1095","conid":"18","topic":"支付系统和金融架构演进（厂商共建）","title":"广发证券新一代行情系统架构与优化","introduce":"<p>行情是券商系统中使用频率最高的功能，其服务质量直接影响用户体验。错误、延迟的行情数据将给投资者造成损失，给公司带来法律风险。同时万级别的标的物，数十种衍生数据，千万级更新订阅，给行情系统带来极大压力。<\/p>\n\n<p>我们从 0 开始，使用 Golang 设计实现了一套十万级更新，百万 QPS 查询，百亿级日推送，高可用低延迟的行情系统。本次分享，将为大家介绍数字背后的架构思路与理念，看接入、推送、业务三大系统如何独立架构，又如何接续整合，以及周边生态的建立。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>行情系统设计的难点：性能与数据可用性<\/li>\n\t<li>数据抽象与分层<\/li>\n\t<li>数据存储与业务<\/li>\n\t<li>高效推送<\/li>\n\t<li>高性能接入层设计与 Golang<\/li>\n\t<li>系统整合<\/li>\n\t<li>协议、缓存与优化<\/li>\n\t<li>生态系统建设<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解行情系统的难点特点、架构与设计<\/li>\n\t<li>了解专用业务优化方法<\/li>\n\t<li>了解 Golang 的落地与优化<\/li>\n\t<li>了解容器、微服务的模式与生态体系<\/li>\n<\/ul>\n","guestName":"王飞","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5955ecbb97170.png","guestIntroduce":"0","beginTime":"1499416500","endTime":"1499419200","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1498803658","updatedAt":"1498803658","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"广发证券","guestPosition":"架构师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/940\/595f4e369fba2.pdf\/18"}],"producter_name":"曹倩芸","topic_id":"193","order":"29","topic_description":"<p>金融业务有其自身特点，安全，稳定是重中之中。如何在保证这些特点的基础上打造适合的金融框架？架构中如何针对系统的健壮性和可维护性进行合理设计，做到高效运维与监控？是很多金融架构创新和演进过程中面临的最大难题。另一方面，随着区块链等一批金融科技创新技术的衍生，其技术实践、未来架构模式、如何与传统网络服务相结合，也一度成为互金领域新的价值探索点。<\/p>\n\n<p>本专题将邀请一线技术人前来分享他们在金融和支付架构演进方面的实践经验，内容聚焦于：支付系统如何从单一功能到完整体系、从臃肿单体PHP演变为高性能高可靠可伸缩的分布式服务架构？金融系统架构的业务特点及架构设计难点，监控与运维有哪些新方法？公有区块链的技术实践以及区块链的未来新架构模式等。<\/p>\n","introduce":"金融业务有其自身特点，安全，稳定是重中之中。如何在保证这些特点的基础上打造适合的金融框架？架构中如何针对系统的健壮性和可维护性进行合理设计，做到高效运维与监控？是很多金融架构创新和演进过程中面临的最大难题。另一方面，随着区块链等一批金融科技创新技术的衍生，其技术实践、未来架构模式、如何与传统网络服务相结合，也一度成为互金领域新的价值探索点。","guestavatar":"http:\/\/static.geekbang.org\/ck\/5951f8f309054.png","company":"InfoQ","position":"策划编辑"},{"topic":"研发团队建设和工程文化","content":[{"id":"840","conid":"18","topic":"研发团队建设和工程文化","title":"数字化平台战略下的IT组织和人才战略 ","introduce":"<p>在数字化转型的过程中，越来越多的企业开始意识到平台思维是通过技术解锁核心业务能力的关键。<\/p>\n\n<p>传统企业在数字化转型过程中，虽然背负着包括遗留系统在内的历史包袱，但历史的资产包括数据、品牌和客户关系也都是转型中的竞争优势。<\/p>\n\n<p>通过数字化平台战略，围绕历史优势资产构建一个生态系统，释放内部创新活力，促进外部生态合作，是企业数字化转型的不二选择。<\/p>\n\n<p>企业的 IT 组织如何在数字化平台战略下构建新的组织结构、跨部门协同模式、人员能力建设和转型期的团队文化建设，是每位 CIO、CTO 和研发主管面临的紧迫课题。<\/p>\n\n<p>本次演讲将围绕数字化平台战略及相应的 IT 组织和人力变更展开。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>企业数字化转型的挑战和机遇；<\/li>\n\t<li>平台思维和数字化平台战略；<\/li>\n\t<li>数字化平台战略下企业 IT 的组织和人力变革；<\/li>\n\t<li>案例分享<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>传统企业的数字化转型平台战略分解；<\/li>\n\t<li>企业生态系统、平台战略下的 IT 组织形式和团队建设。<\/li>\n<\/ul>\n","guestName":"施韵涛","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7986901dc4.png","guestIntroduce":"0","beginTime":"1499405400","endTime":"1499408100","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1490886723","updatedAt":"1490886723","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"ThoughtWorks","guestPosition":"中国区交付服务总监","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/921\/595f23ac9ce4f.pdf\/18"},{"id":"915","conid":"18","topic":"研发团队建设和工程文化","title":"建一支分布式的远程团队","introduce":"<p>过去一年我在创业，建立了一支分布式的远程工作团队，8人技术人员分布在5个城市，整个团队在一种完全自由和自治的过程中开发出了公司的核心产品，在整个过程中，我和我的团队收获了远程工作的经验和心得。<\/p>\n\n<p>我们发现，远程工作完全自己管理自己，自己驱动自己的方式，并不比在一起办公的效率要低，相反，远程工作对于规范软件团队的工作习惯非常非常有帮助。<\/p>\n\n<p>比如项目的文档、设计和代码评审、自动化的流程、以及相关的远程工作的支持工具，都会比在一起工作的效率要高，质量要好。我想通过个话题和大家分享一下分布式远程工作的心得。<\/p>\n","guestName":"左耳朵耗子","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7528d95807.png","guestIntroduce":"0","beginTime":"1499409000","endTime":"1499411700","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1492581129","updatedAt":"1492581129","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"MegaEase","guestPosition":"创始人","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/930\/595f337d54e3b.pdf\/18"},{"id":"807","conid":"18","topic":"研发团队建设和工程文化","title":"独角兽团队的持续迭代","introduce":"<p>结合创业公司业务发展不同阶段，如何做好技术的选型和演进？技术团队应该怎么保持快速的迭代？技术文化和氛围应该怎样打造？如何形成技术的影响力和战斗力？<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>初阶：建立流程规范及机制；<\/li>\n\t<li>进阶：团队梯队搭建以及组织分工；<\/li>\n\t<li>高阶：团队文化建设，保持创业公司的战斗力。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>创业公司如何快速搭建团队和做好文化建设；<\/li>\n\t<li>如何确保在业务快速发展过程中做好技术选型和演进；<\/li>\n\t<li>如何保持创业公司的战斗力。<\/li>\n<\/ul>\n","guestName":"徐裕键","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7635a61b05.png","guestIntroduce":"0","beginTime":"1499412900","endTime":"1499415600","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d25496e02e3.png","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1490010389","updatedAt":"1490010389","status":"0","days":"07","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d25496e02e3.png","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"贝贝网","guestPosition":"合伙人&研发副总裁","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/936\/595f443ecade9.pdf\/18"},{"id":"751","conid":"18","topic":"研发团队建设和工程文化","title":"Move Fast and Break Things: Engineering at Facebook","introduce":"<p>A few billion likes and a few hundred million photos uploaded a day. A hundred or so petabytes of queryable data and a handful of power hungry datacenters. And it&#39;s all humming while code is being written and deployed twice daily without our users noticing.<\/p>\n\n<p>This talk will explore the &quot;move fast&quot; side of Facebook&#39;s software engineering culture: development process, organizational structure and the vast amounts of tooling we use to make sure we don&#39;t screw up.<\/p>\n\n<p>We&#39;ll also dig into how we &quot;ship things&quot;: release process, A\/B testing, gate keepers, test infrastructure and so on. You&#39;ll walk away with a prospective of how we build software at scale, for scale.<\/p>\n\n<p>一天几十亿的点赞，一天几亿的照片上传，上百个 Perabytes 的可搜索数据，数个大型数据中心，在所有这些安静高效的运转的背后，新代码在不断大量的产生，每天两次上线，平稳无误，这就是 Facebook 的软件工程！而在这背后究竟是什么样的工程文化呢？<\/p>\n\n<p>这次分享将向你展示 Facebook &ldquo;moving fast&rdquo; 的工程文化。Facebook 以快著称，快体现在我们的开发流程上，体现在我们的组织结构上，体现在我们众多的工具支持上，有了它们，我们才可以做到快而不错。<\/p>\n\n<p>我们还将深入讨论我们是如何 &ldquo;ship things&rdquo;，包括我们的发布流程、A\/B 测试、Gatekeeper 系统、测试系统等等。相信听完我们的分享，您将会对 Facebook 是如何进行大型软件开发有所了解。<\/p>\n","guestName":"Joel Pobar","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f756d726843.png","guestIntroduce":"0","beginTime":"1499416500","endTime":"1499419200","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d25446e15af.png","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1489371362","updatedAt":"1489371362","status":"-1","days":"07","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d25446e15af.png","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"Facebook","guestPosition":"Engineering Director","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/938\/595f4a48037bd.pdf\/18"}],"producter_name":"康德胜","topic_id":"96","order":"24","topic_description":"<p>当技术逐渐成为企业的核心竞争力，如何打造一个高效高质灵活的研发团队以及富有朝气的企业文化，是每一个技术管理者最关注的领域。<\/p>\n\n<p>怎样才能做好真正的敏捷项目管理？团队绩效是用 KPI 还是 OKR？如何建设工程师文化提高开发效率？<\/p>\n\n<p>产品经理如何同技术开发和谐共处？怎样利用微服务技术架构提升团队的灵活性？中等规模的公司技术人员是归 CTO 管理还是直接向业务线汇报？<\/p>\n\n<p>技术人才该如何培养？什么样的组织架构更能把 DevOps 以及 CICD 应用好？<\/p>\n\n<p>我们希望研发管理者能够在这个论坛上得到这些问题的答案。&nbsp;<\/p>\n","introduce":"当技术逐渐成为企业的核心竞争力，如何打造一个高效高质灵活的研发团队以及富有朝气的企业文化，是每一个技术管理者最关注的领域，我们希望研发管理者能够在这个论坛上得到这些问题的答案。 ","guestavatar":"http:\/\/static.geekbang.org\/ck\/58bd26bd367c6.png","company":"万惠集团","position":"副总裁&CTO"},{"topic":"移动端架构实践与优化","content":[{"id":"841","conid":"18","topic":"移动端架构实践与优化","title":"微信 Android 模块化架构重构实践","introduce":"<p>过去半年多的时间里，为了重塑微信的模块化，实现轻量微信业务内核支持微信硬件设备的需要，我们对微信 Android 终端架构进行了一次新的重构。<\/p>\n\n<p>经过重构，微信实现了相当轻量的业务内核，能在短时间内开发具备核心功能的微信原型。同时面对微信敏捷迭代的开发方式和内部关联复杂的业务特点，我们重新构建了微信的模块化框架，用于解决以前模块边界，模块依赖，模块通信，及启动管理等面临的实际问题。<\/p>\n\n<p>更进一步，我们借此推动模块的接口化、文档化和可测试化，及开发模式上的转变。<\/p>\n\n<p>这次重构实践具有很多工程化普遍适用的价值，我们希望能借此机会分享其中的设计理念、设计方式，与业界同行探讨交流。<\/p>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>架构如何被&ldquo;腐蚀&rdquo;的，怎么避免架构变质；<\/li>\n\t<li>模块化架构设计需要考虑和解决的各种问题，及微信架构工程化实践所具有的普遍适用价值；<\/li>\n\t<li>终端架构设计和工作流配合的思考；<\/li>\n\t<li>其他业界方案的比较，在工程化最佳的角度上利弊权衡。<\/li>\n<\/ul>\n","guestName":"郭锐","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f79790eb0c7.png","guestIntroduce":"0","beginTime":"1499405400","endTime":"1499408100","address":"大宴会厅2","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58dd34d5a7788.png","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1490891198","updatedAt":"1490891198","status":"0","days":"07","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58dd34d5a7788.png","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"腾讯","guestPosition":"微信Android功能开发组高级工程师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/927\/595f308aae725.pdf\/18"},{"id":"824","conid":"18","topic":"移动端架构实践与优化","title":"360手机卫士性能提升攻略","introduce":"<p>360手机卫士基于插件化开发框架，拥有90多个插件，它继承了插件化的快速发布及多团队并行开发的优势，同时面临着各种性能方面的挑战。如此一款代码量及用户量巨大的超级 APP，性能方面的一点提升都会使大量用户受益。<\/p>\n\n<p>我们 APM 团队针对性能问题，从过度绘制、Activity 启动时间、传感器、耗电等多维度都做了专项优化，使卫士的功耗在各个场景都有了30%以上的优化，用户 UI 体验也有了大幅度提升。<\/p>\n\n<p>我们通过完善移动端性能管理体系，包括性能规范、代码红线、APM SDK、诊断上报、自动化测试、性能测试、性能数据后台等，及时发现并解决卫士性能问题，持续提升卫士性能，提升用户的产品体验。<\/p>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>一款超级 APP 遇到的性能难题；<\/li>\n\t<li>从哪些方面来分析和优化 APP性能；<\/li>\n\t<li>如何持续提升一款 APP性能；<\/li>\n\t<li>一些性能借鉴案例。<\/li>\n<\/ul>\n","guestName":"卜云涛","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7640c13234.png","guestIntroduce":"0","beginTime":"1499409000","endTime":"1499411700","address":"大宴会厅2","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58dd368419c1b.png","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1490348446","updatedAt":"1490348446","status":"-1","days":"07","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58dd368419c1b.png","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"奇虎360","guestPosition":"手卫APM负责人","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/926\/595f2f7de1d5f.pdf\/18"},{"id":"918","conid":"18","topic":"移动端架构实践与优化","title":"Atlas: 手机淘宝 Android 架构实践","introduce":"<p>Atlas&nbsp;是伴随着手机淘宝的不断发展而衍生出来的一个运行于 Android 系统上的一个 Native 容器化框架，也叫动态组件化(Dynamic Bundle)框架。<\/p>\n\n<p>它主要提供了解耦化、组件化、动态性的支持，解决了团队协作的问题，解决了业务动态更新的问题。目前淘宝系80%以上的 App 都在使用，价值和稳定性也已经得到了充分的证明。<\/p>\n\n<p>从2014年被手机淘宝团队首次提出，到2017年初千锤百炼的 Atlas 终于开源。本次演讲将从 Altas 的设计原理入手，深入剖析客户端容器化框架的解决方案，并深入浅出的带来应用实践。<\/p>\n\n<p>拥抱 Atlas，复杂的事情可以变简单；<\/p>\n\n<p>返璞归真，小小的屏幕足以释放我们的想象；<\/p>\n\n<p>回归初心，带您找回移动开发最初的简单和快乐。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>从手淘面临问题阐述 Atlas 的产生；<\/li>\n\t<li>剖析 Atlas&nbsp;技术原理；<\/li>\n\t<li>分析动态加载原理。<\/li>\n<\/ul>\n","guestName":"胡文江（白衣）","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/590aef8a2ce7d.png","guestIntroduce":"0","beginTime":"1499412900","endTime":"1499415600","address":"大宴会厅2","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1492588719","updatedAt":"1492588719","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"阿里巴巴","guestPosition":"高级无线技术专家","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/935\/595f42c5d382e.pdf\/18"},{"id":"1039","conid":"18","topic":"移动端架构实践与优化","title":"电商3D模型数据平台","introduce":"<p>最近几年 VR\/AR 席卷了各行各业，尤其对电商这一线上平台来说，VR\/AR 由于能够补全或增强用户使用商品的体验，拥有无限的想象空间，各大电商都开始布局 VR\/AR。但无论是 VR 还是 AR，都依赖于 3D 化的商品数据库。<\/p>\n\n<p>将以 2D 图片为主的商品数据库 3D 化将是一个长期而艰巨的工作，所以长远来看，构建一个集商品 3D 模型制作、预览、上线及展示的平台是整个 VR\/AR 战略最基础的一环。而搭建这一平台的关键技术是流程的梳理以及渲染引擎。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>背景<\/li>\n\t<li>平台架构<\/li>\n\t<li>模型制作<\/li>\n\t<li>模型上线<\/li>\n\t<li>模型展示<\/li>\n\t<li>瓶颈与展望<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解如何构建3D数据平台<\/li>\n\t<li>了解如何打造轻量级渲染引擎<\/li>\n<\/ul>\n","guestName":"刘享军","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/593135b7e8f64.png","guestIntroduce":"0","beginTime":"1499416500","endTime":"1499419200","address":"大宴会厅2","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1496397769","updatedAt":"1496397769","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"京东","guestPosition":"平台研发部-VRAR业务部算法架构师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/942\/595f52439a6aa.pdf\/18"}],"producter_name":"张渝","topic_id":"92","order":"22","topic_description":"<p>微信小程序、Google Progressive Web Apps 以及各种移动应用的新形式逐渐受重视。<\/p>\n\n<p>集成大量业务的超级 App 开始遇到遇各种性能问题，牵一发动全身情况下该从哪些方面来分析和优化？<\/p>\n\n<p>在新技术潮流和谨慎迭代现有架构下我们如何选择，又有哪些成功的可供借鉴的技术案例？<\/p>\n","introduce":"微信小程序、Google Progressive Web Apps 以及各种移动应用的新形式在移动领域的分量越来越重，在这些新技术新潮流下，有哪些值得借鉴的技术变化呢？","guestavatar":"http:\/\/static.geekbang.org\/ck\/58d481a5867a3.png","company":"腾讯","position":"增值产品部开发总监"},{"topic":"运维新挑战","content":[{"id":"887","conid":"18","topic":"运维新挑战","title":"小米监控实践之路","introduce":"<p>从2011年到2014年，小米经历了快速成长，随着机器数量增加，IDC 环境越来越复杂，原有的监控系统已经无法满足需求，我们因此设计小米监控，追求易用、高可用，力求全程自动化。<\/p>\n\n<p>小米每天都有百余台服务器交付或下线，上千次的业务变更，如果这些都靠 SRE 进行人工维护，将会消耗大量人力，管理效率低下的同时也容易出现遗漏、配错等问题。<\/p>\n\n<p>本次分享将向大家介绍，小米监控系统是如何融入自动化体系，与各个系统协作，在业务运维的全生命周期中进行全自动管理。<\/p>\n\n<h2>听众收益<\/h2>\n\n<ul>\n\t<li>小米监控曾经面临的困境；<\/li>\n\t<li>小米监控设计思路和原则；<\/li>\n\t<li>小米监控如何实现全自动管理。<\/li>\n<\/ul>\n","guestName":"陈帅","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f79b5a05671.png","guestIntroduce":"0","beginTime":"1499405400","endTime":"1499408100","address":"爱晚亭","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1491804427","updatedAt":"1491804427","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"小米","guestPosition":"云平台工程师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/924\/595f2d5de2fb1.pdf\/18"},{"id":"750","conid":"18","topic":"运维新挑战","title":"腾讯海量监控包袱与创新","introduce":"<p>腾讯社交业务规模庞大，历史悠久，架构复杂。从运维的全局角度来看，无论从运维技术还是监控难度都很大。<\/p>\n\n<p>传统的监控手段和思想已经无法应对如此海量的场景，腾讯社交网络运营部历经十年的建设，在运维监控领域经过了多个建设阶段。<\/p>\n\n<p>近几年通过创新的方法引入了多种技术手段并实践落地，将监控技术带入一个新的运维高度，本次将主要分享四个创新技术点。<\/p>\n\n<h2>创新技术点<\/h2>\n\n<ul>\n\t<li>通过业务架构结合数据链路计算选路等方式来发掘告警的故障根源，提升故障定位效率；<\/li>\n\t<li>通过建立生命指标来做告警分级运营管理和生态建设的运营方法论；<\/li>\n\t<li>通过海量数据搜集和全链路数据挖掘来深入定位问题，并深化数据价值；<\/li>\n\t<li>通过机器训练来进行自适应告警优化。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>分享将聚焦在腾讯运维近几年在运维监控体系建设方面的思考和创新方法与实践，希望能帮助听众了解；<\/li>\n\t<li>海量运维中监控体系的产品演进、技术更新步伐、技术短板出现时期、已有的创新方法实践效果、各种创新方法印证过的技术坑，及未来探索的思路；<\/li>\n\t<li>希望通过腾讯的主流技术案例和项目经验，与同行交流和探讨，扬长避短共同进步。<\/li>\n<\/ul>\n","guestName":"聂鑫","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58c5fb0c7ac89.png","guestIntroduce":"0","beginTime":"1499409000","endTime":"1499411700","address":"爱晚亭","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d2530078b0d.png","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1489370158","updatedAt":"1489370158","status":"0","days":"07","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d2530078b0d.png","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"腾讯","guestPosition":"运维总监","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/987\/597eccb873193.pdf\/18"},{"id":"912","conid":"18","topic":"运维新挑战","title":"携程容器云智能运维","introduce":"<p>携程从2016年开始开发容器云，并且在生产环境上线了基于容器的持续交付系统。容器的带来的标准化，轻量隔离，为应用的快速交付提供了可能，但是也对原有的运维体系带来了挑战。<\/p>\n\n<p>在容器云的落地过程中，我们引入了新的日志，监控方式，并且通过 StackStorm，chatbot 的方式让 DevOps 的理念很好的落地。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>物理机-&gt;虚拟机-&gt;容器，基础设施的变迁史，带来的机遇与挑战；<\/li>\n\t<li>基于容器的持续交付体系介绍；<\/li>\n\t<li>携程容器云建设细节：镜像、网络、调度、监控、日志、计费等；<\/li>\n\t<li>基于 StackStorm&nbsp;的&nbsp;DevOps 实践<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>我们是如何在容器云中进行监控和日志管理的；<\/li>\n\t<li>从虚拟机时代到容器时代，运维方式和手段发生了变化，我们是如何进行平衡的。<\/li>\n<\/ul>\n","guestName":"蔡峰","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7a471ae7fe.png","guestIntroduce":"0","beginTime":"1499412900","endTime":"1499415600","address":"爱晚亭","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1492437033","updatedAt":"1492437033","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"携程","guestPosition":"技术保障中心云平台研发总监","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/933\/595f3c0ed4d0d.pdf\/18"},{"id":"999","conid":"18","topic":"运维新挑战","title":"全链路稳定性背后的数字化支撑：阿里巴巴鹰眼技术解密","introduce":"<p>阿里的电商业务系统基于微服务框架构建，其变更频繁，业务变化快，系统依赖关系错综复杂。基于全链路的稳定性保障、容量规划、性能优化、压测演练、业务故障排查定位等运维手段也逐步系统化、常态化。<\/p>\n\n<p>从2010年的第一代&ldquo;鹰眼&rdquo;系统诞生以来，阿里的全链路监控系统不断升级、演进与创新，为 DevOps 和 SRE 在阿里的实际落地提供了行之有效的数字化支撑。<\/p>\n\n<p>最新一代的阿里全链路监控系统鹰眼3.0，同时将基础设施层、分布式应用层、业务逻辑层与客户端层进行了全链路跟踪；技术层面，鹰眼3.0日均处理万亿级别的分布式调用链数据，针对海量实时监控的痛点，对底层的流计算、多维时序指标与事件存储体系等进行了大量优化，同时引入了时序检测、根因分析、业务链路特征等技术，将问题发现与定位由被动转为主动。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>分布式链路追踪技术原理、基础功能以及在阿里巴巴的使用场景。<\/li>\n\t<li>阿里全链路监控平台技术架构演进，计算引擎、存储技术细节介绍，教你如何构建一个大规模监控系统。<\/li>\n\t<li>精细化监控 - 监控系统如何满足多变的业务监控场景。<\/li>\n\t<li>由被动转为主动 - 如何&ldquo;玩&rdquo;好监控数据。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>从监控指标的收集与清洗技术、流计算平台、海量运维指标与事件的存储技术到预警技术的全面剖析，听众可以了解到建设大型IT监控系统时的架构、技术痛点与解决方案；<\/li>\n\t<li>了解如何体系化构建自己的全链路监控平台；<\/li>\n\t<li>基于应用拓扑结构、时间序列数据挖掘的异常检测与离群点检测、基于海量业务非结构化数据的智能诊断，了解阿里监控体系的技术创新与探索。<\/li>\n<\/ul>\n","guestName":"周小帆","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/591bfa7d05c06.png","guestIntroduce":"0","beginTime":"1499416500","endTime":"1499419200","address":"爱晚亭","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1495008587","updatedAt":"1495008587","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"阿里巴巴","guestPosition":"高级技术专家","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/939\/595f4cdcb9d52.pdf\/18"}],"producter_name":"赵成（谦益）","topic_id":"80","order":"27","topic_description":"<p>在云计算遍及业界的趋势下，以及 DevOps 和 SRE 等先进运维理念的强势助推，运维已然成为驱动各大公司研发运维流程和理念变革的关键角色，如持续集成和发布、场景化的运维自动化、智能监控等理念的落地执行。<\/p>\n\n<p>同时，运维所从事的工作角色定位也在悄然地发生着变化，从原来的末端被动响应，逐步转向技术产品、技术运营和平台建设者的角色。<\/p>\n\n<p>我们很欣喜地看到，运维已经慢慢承担起了稳定性保障、流程效率改进、性能优化、用户体验提升以及成本控制等关键职责，但更高的要求必然带来新的挑战和机遇，我们将如何应对？<\/p>\n\n<p>本专题将会邀请一线运维专家，在基于容器的持续集成和发布、智能监控和故障自愈、成本和性能优化几个方向上，分享他们的实践和思考，看看专家们在技术高速发展的趋势下，是如何应对这些新的挑战的。<\/p>\n\n<p>&nbsp;<\/p>\n","introduce":"本专题将会邀请一线运维专家，在基于容器的持续集成和发布、智能监控和故障自愈、成本和性能优化几个方向上分享实践和思考，看看专家们在技术高速发展的趋势下是如何应对这些新的挑战的。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58b78a17d7833.png","company":"美丽联合集团","position":"运维经理"},{"topic":"大数据框架","content":[{"id":"900","conid":"18","topic":"大数据框架","title":"Secret Kung Fu of Massive Scale Stream Processing with Apache Samza","introduce":"<p>Event processing is a race against time: a race where seconds or even milliseconds provide greater relevancy and accuracy of the results than hours or days.<\/p>\n\n<p>To lead this race, we&#39;ve been running 400+ Samza applications reliably in production over the past 5 years at LinkedIn, processing over 1 trillion events each day. So, what&#39;s the secret ingredients behind it?<\/p>\n\n<p>In this talk we will inspect some of them:<\/p>\n\n<p>&nbsp; a) a fluent API that allows the user to focus on the processing logic without worrying about the execution details;<\/p>\n\n<p>&nbsp; b) versatile deployment models that allows us to run Samza applications in Yarn cluster, as well as clusters like AWS&#39;EC2;<\/p>\n\n<p>&nbsp; c) durable local state that can scale large stateful applications with ease;<\/p>\n\n<p>&nbsp; d) asynchronous processing that enables remote data I\/O to match the throughput of event consumption.<\/p>\n\n<p>Finally, we will also explore patterns that allow us to run the same application in both nearline and offline.<\/p>\n\n<p>参考翻译：<\/p>\n\n<p>事件处理是场争分夺秒的比赛，需要在秒级甚至毫秒级内提供更好的关联性和精确性。<\/p>\n\n<p>为了赢得这场比赛，在过去领英5年的生产环境中，我们已经稳定地运行了400+Samza 应用程序，每天处理超过万亿事件，在这一切背后，领英拥有怎样的秘籍呢？<\/p>\n\n<p>在这次分享中，我们将观察 Samza&nbsp;的以下特性：<\/p>\n\n<p>a)一个流处理API，允许用户专注于处理逻辑，而不用担心执行细节；<\/p>\n\n<p>b)灵活的部署模式，我们能够在Yarn集群或AWS&#39;EC2集群中运行Samza应用；<\/p>\n\n<p>c)在本地持久化中间状态，从而轻松应对大规模的有状态应用的扩展<\/p>\n\n<p>d)支持异步处理，因此能够允许远程数据I\/O支持整个事件消费产生的吞吐量。<\/p>\n\n<p>e)最后，我们还会探索Samza作为一个平台如何兼容流处理和批处理两种模式。<\/p>\n\n<p>&nbsp;<\/p>\n","guestName":"刘新宇","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f79fc60fc1d.png","guestIntroduce":"0","beginTime":"1499405400","endTime":"1499408100","address":"大宴会厅1","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1491939946","updatedAt":"1491939946","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"LinkedIn","guestPosition":"Staff Software Engineer","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/931\/595f357c2835d.pdf\/18"},{"id":"1076","conid":"18","topic":"大数据框架","title":"Sloth: 网易流计算服务化平台架构实践","introduce":"<p>网易历经了20年的发展，各业务线沉淀了丰富的数据。大数据平台在满足各业务线数据存储、计算的任务，同时承担着整合集团数据，为客户提供全方位的大数据服务。<\/p>\n\n<p>在满足业务需求的过程中，对 Hadoop 各组件进行了定制优化（如分时调度，权限控制，元数据管理，Spark 多租户，Spark 高可用，流计算相关）。<\/p>\n\n<p>同时，围绕大数据底层技术平台也孵化出了猛犸大数据开发平台，网易有数敏捷数据分析平台等产品。本次演讲重点介绍网易基于 Flink 研发的流计算服务化平台，以 SQL 为主要开发方式，支持 DDL，支持丰富的 DML，如 Stream join, Window，支持 UDF，提供丰富的 connector。为网易各业务线提供一站式的流计算解决方案。&nbsp;<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>&nbsp;Hadoop in Netease&nbsp;<\/li>\n<\/ul>\n\n<p style=\"margin-left:40px\">a.网易数据平台架构&nbsp;<\/p>\n\n<p style=\"margin-left:40px\">b.数据平台如何去支持各种产品线&nbsp;<\/p>\n\n<p style=\"margin-left:40px\">c.Netease Hadoop 创新与改进&nbsp;<\/p>\n\n<ul>\n\t<li>增量计算模型&nbsp;<\/li>\n<\/ul>\n\n<p style=\"margin-left:40px\">a.什么是增量计算&nbsp;<\/p>\n\n<p style=\"margin-left:40px\">b.增量计算与流计算的关系&nbsp;<\/p>\n\n<p style=\"margin-left:40px\">c.为什么需要使用增量计算的方式去处理流计算&nbsp;<\/p>\n\n<p style=\"margin-left:40px\">d.Sloth 是如何 build 增量计算模型&nbsp;<\/p>\n\n<ul>\n\t<li>流计算服务化平台&nbsp;\n\t<p style=\"margin-left:2px\">a.业务背景&nbsp;<\/p>\n\n\t<p style=\"margin-left:2px\">b.Sloth 整体架构&nbsp;<\/p>\n\n\t<p style=\"margin-left:2px\">c.Streaming SQL 上是创新&nbsp;<\/p>\n\n\t<ol style=\"list-style-type:lower-roman\">\n\t\t<li>DDL&nbsp;<\/li>\n\t\t<li>任务级联&nbsp;<\/li>\n\t\t<li>UDF&nbsp;<\/li>\n\t\t<li>复杂 operator，如window, join<\/li>\n\t<\/ol>\n\t<\/li>\n\t<li>未来规划&nbsp;<\/li>\n<\/ul>\n\n<p style=\"margin-left:40px\">a.统一计算模型，Apache Beam&nbsp;<\/p>\n\n<p style=\"margin-left:40px\">b.离线在线混部&nbsp;<\/p>\n\n<p style=\"margin-left:40px\">c.新硬件的应用，FPGA，GPU&nbsp;<\/p>\n\n<p style=\"margin-left:40px\">d.TensorFlow on Hadoop&nbsp;<\/p>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解大规模流计算服务平台如何去工作&nbsp;<\/li>\n\t<li>了解如果去统一 Streaming 与 Batch 模型&nbsp;<\/li>\n\t<li>了解网易是如何根据自己的业务场景进行大数据数据平台实践的&nbsp;<\/li>\n<\/ul>\n","guestName":"金晓军","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5949f81fe470f.png","guestIntroduce":"0","beginTime":"1499409000","endTime":"1499411700","address":"大宴会厅1","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1498023199","updatedAt":"1498023199","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"网易","guestPosition":"数据科学中心技术负责人&首席架构师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/944\/595f5e8b6b90b.pdf\/18"},{"id":"1014","conid":"18","topic":"大数据框架","title":"Airbnb 数据仓库架构实践","introduce":"<p>Airbnb 目前的大数据平台支撑了整个公司对数据收集和处理的需求，我们需要应对每天100亿条新增日志信息，350TB 新增数据，来自公司所有部门的10多万次大数据分析查询，而且还在快速增长中。<\/p>\n\n<p>本演讲从数据处理需求、性能、系统容错以及服务等级协议(SLA)等各方面入手，重点讲解我们遇到的挑战和随之而生的平台解决方案，介绍我们研发的 ReAir 跨机群同步服务，AirStream 流处理平台以及实时数据注入和查询平台等。<\/p>\n\n<h2>演讲大纲<\/h2>\n\n<ul>\n\t<li>过去的数据平台：\n\t<p style=\"margin-left:2px\">a. 规模<\/p>\n\n\t<p style=\"margin-left:2px\">b. 存在的问题<\/p>\n\t<\/li>\n\t<li>数据处理需求：\n\t<p style=\"margin-left:2px\">a. 数据流水线批处理<\/p>\n\n\t<p style=\"margin-left:2px\">b. 即席(ad-hoc)查询<\/p>\n\t<\/li>\n\t<li>Airbnb 的数据仓库：\n\t<p style=\"margin-left:2px\">a. 多机群架构 (Gold\/Silver)<\/p>\n\n\t<p style=\"margin-left:2px\">b. 低延时跨机群数据同步 - ReAir<\/p>\n\n\t<p style=\"margin-left:2px\">c. 采用 Spark Streaming\/HBase 的 AirStream 流数据处理平台<\/p>\n\n\t<p style=\"margin-left:2px\">d. 实时数据注入(data ingestion)<\/p>\n\n\t<p style=\"margin-left:2px\">e. 实现 Presto\/HBase connector 进行实时数据查询<\/p>\n\n\t<p style=\"margin-left:2px\">f. HDFS + S3 混合存储架构<\/p>\n\t<\/li>\n\t<li>数据仓库的未来优化：\n\t<p style=\"margin-left:2px\">a. 加速 S3 元数据访问<\/p>\n\n\t<p style=\"margin-left:2px\">b. 联合集群 (Federated Clusters)<\/p>\n\t<\/li>\n<\/ul>\n","guestName":"曾洪博","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5925203a30cd4.png","guestIntroduce":"0","beginTime":"1499412900","endTime":"1499415600","address":"大宴会厅1","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1495605721","updatedAt":"1495605721","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"Airbnb","guestPosition":"资深工程师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/934\/595f3e893415d.pdf\/18"},{"id":"1038","conid":"18","topic":"大数据框架","title":"美团点评数据平台融合实践","introduce":"<p>美团点评两家公司合并以来，经历了组织架构融合、线上业务融合、数据与基础设施融合的三个大阶段。<\/p>\n\n<p>本次分享主要介绍第三阶段，美团点评整个数据平台融合的项目推进经验，以及相关技术突破，期望能对于有大的数据平台技术产品重构或底层技术架构调整需求的架构师进行参考。<\/p>\n\n<p>平台项目分为三个阶段：<\/p>\n\n<ul>\n\t<li>数据流打通；<\/li>\n\t<li>离线集群透明迁移融合；<\/li>\n\t<li>工具链产品融合与业务任务切分。<\/li>\n<\/ul>\n\n<p>融合之前，我们有异地双机房共4k节点，涉及万级别任务迁移与自动改写，整个融合过程持续提供数据生产与应用服务，要求对业务透明。<\/p>\n\n<p>本次分享除了对于项目整体计划与实施的讨论，也会会涵盖 Hadoop 集群认证打通、Hadoop 多机房架构改造、大面积 SQL 任务重构的自动化方法等核心技术点的详细架构介绍。<\/p>\n\n<p>其中，Hadoop 集群认证打通我们使用 Kerberos 跨域认证方案，Hadoop 多机房方案我们通过新增 Zone Tunnel 模块，操作 HDFS&nbsp;文件块的分布，做到了集群的平滑迁移。<\/p>\n\n<p>在 SQL 任务自动重构的场景下, 我们自研了一个 Hive 表映射的方案，使得任务的修改可以并行化，无需双写，也无需分层迁移，大大的提高了重构效率。<\/p>\n","guestName":"谢语宸","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5931320173f33.png","guestIntroduce":"0","beginTime":"1499416500","endTime":"1499419200","address":"大宴会厅1","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1496396257","updatedAt":"1496396257","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"美团点评","guestPosition":"离线大数据平台方向技术负责人","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/943\/595f58d3c07e9.pdf\/18"}],"producter_name":"严岩","topic_id":"95","order":"21","topic_description":"<p>大数据框架从提起到流行不足20年，期间产生耳熟能详的 Hadoop、Spark 和 Storm 等框架。<\/p>\n\n<p>如何利用大数据框架实现数据挖掘、预测分析、语义引擎和数据质量管理等功能？<\/p>\n\n<p>如何选择并落地合适的大数据框架？让我们从不同的企业应用中探讨各种大数据框架的应用背景、实践以及教训。<\/p>\n","introduce":"常见的大数据框架有 Hadoop、Spark、Storm 等，这些框架之争没有优劣，只有合适与否，本专题从不同的企业应用中，探讨各种大数据框架的应用背景、实践以及教训。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58bd25ab6671b.png","company":"LinkedIn","position":"Senior Software Engineer"},{"topic":"推荐系统架构实践","content":[{"id":"902","conid":"18","topic":"推荐系统架构实践","title":"大数据驱动业务增长：百度外卖商业智能推荐系统实践","introduce":"<p>O2O 平台上的交易服务和用户行为提供了海量数据，从中挖掘出的商户画像和用户画像有助于更深刻地理解商业本质。根据这些数据分析和挖掘，除了锦上添花地提升用户消费体验，如何雪中送炭地驱动整个平台的业务增长，是一个更加重要的问题。<\/p>\n\n<p>本次分享会结合百度外卖商业智能推荐系统（BI RecSys）的构建过程，向大家剖析其中的常见模型、主要设计原理和核心技术，包括在不同发展阶段如何通过建模、预测和反馈评估来平衡平台、商家和用户三者之间的自身价值和利益诉求，促进商业信息更有效率地传输，驱动业务快速健康增长。<\/p>\n\n<p>演讲提纲<\/p>\n\n<ul>\n\t<li>商户画像和用户画像建设；<\/li>\n\t<li>可视化平台与 API 接口输出；<\/li>\n\t<li>根据业务要素提炼应用场景能力；<\/li>\n\t<li>分阶段差异化运用增长工具。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解商业大数据建模的基本过程；<\/li>\n\t<li>了解在商业智能领域针对不同场景构建推荐系统的方法论；<\/li>\n\t<li>了解针对具体业务目标实操大数据增长的经验。<\/li>\n<\/ul>\n","guestName":"蒋凡","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7a142ac57b.png","guestIntroduce":"0","beginTime":"1499405400","endTime":"1499408100","address":"马德里5号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1491941002","updatedAt":"1491941002","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"百度外卖","guestPosition":"技术委员会主席&主任架构师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/923\/595f2c5e10125.pdf\/18"},{"id":"1099","conid":"18","topic":"推荐系统架构实践","title":"TensorFlow在推荐系统中的应用和实现","introduce":"<p>从核心算法上讲，推荐系统是如何经历从最初的协同过滤到如今的深度学习等各种复杂的机器学习模型的演变。<\/p>\n\n<p>本次分享将演示如何使用 Tensorflow 实现各种模型，从简单的浅层神经网络双线性模型，到针对推荐系统稀疏数据场景下的 Amazon 的 DSSTNE，和 Youtube 推荐系统采用的神经网络在 TensorFlow 上的实现。<\/p>\n\n<p>TensorFlow 如何实现 Wide &amp; Deep Learning 的结合。如何结合 Wide Model，也就是传统的基于 Logistic Regression 并且用大量的 Cross-Product 做 Feature，和用 Deep Neural Networks 把大量 Categorial Features 转换成为 Dense Embedding Vector。<\/p>\n\n<p>同时，还将使用 TensorBoard，来可视化物品（items）和元数据（metadata）的潜在空间（latent space）。以及线上和线下的模型评估方法。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>Lookalike 算法实现的全景架构，传统的 CTR 或推荐系统拥有高维特征和稀疏数据，转向深度学习如何高效实现；<\/li>\n\t<li>浅层神经网络双线性模型、Amazon 的 DSSTNE、Youtube 推荐系统采用的神经网络在 TensorFlow 上的实现；<\/li>\n\t<li>如何使用 TensorBoard，来可视化物品（items）和元数据（metadata）的潜在空间（latent space）；<\/li>\n\t<li>TensorFlow 如何实现 Wide &amp; Deep Learning 的结合。&nbsp;<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解深度学习在推荐系统中的应用和架构；<\/li>\n\t<li>了解 TensorFlow 实现常见神经网络的原理以及具体实现；<\/li>\n\t<li>了解如何使用 TensorBoard 等可视化工具来可视化物品和元数据的关系。<\/li>\n<\/ul>\n","guestName":"李嘉璇","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/595a0670a1e22.png","guestIntroduce":"0","beginTime":"1499409000","endTime":"1499411700","address":"马德里5号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1499072302","updatedAt":"1499072302","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"《TensorFlow技术解析与实战》","guestPosition":"作者","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/986\/596b86635c4a2.pdf\/18"},{"id":"972","conid":"18","topic":"推荐系统架构实践","title":"美团外卖原生广告推荐实践 ","introduce":"<p>经过几年的发展，美团外卖的日订单已经突破千万。业务成长到一定阶段后，进行商业变现是一个常见的问题。何如针对业务定制一套合适的系统并进行针对性的算法优化，以及商业系统如何与推荐系统融合，是其中更加重要的问题。 &nbsp;<\/p>\n\n<p>本分享结合美团外卖商业从0到1的发展过程，向大家介绍背后的原生广告的思考，及如何将商业与自然结果进行联合优化，如何在用户推荐体验、商户体验、平台收益之间进行决策，以及面对针对快速增长的业务何如做好技术架构的保障。 &nbsp;<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>业务背景介绍；<\/li>\n\t<li>商业&amp;推荐结果的有机融合；<\/li>\n\t<li>推荐体验&amp;商户效果&amp;平台收益的优化；<\/li>\n\t<li>高增长业务的架构保障。<\/li>\n<\/ul>\n\n<h2>听众收益<\/h2>\n\n<ul>\n\t<li>了解外卖业务特点；<\/li>\n\t<li>了解原生广告实践经验；<\/li>\n\t<li>了解广告架构实践经验。<\/li>\n<\/ul>\n","guestName":"王兴星","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/591958ac16963.png","guestIntroduce":"0","beginTime":"1499412900","endTime":"1499415600","address":"马德里5号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1494833723","updatedAt":"1494833723","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"美团外卖","guestPosition":"商业技术负责人&高级专家","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/928\/595f309bc9144.pdf\/18"},{"id":"919","conid":"18","topic":"推荐系统架构实践","title":"阿里推荐平台实战分享","introduce":"<p>个性化推荐业务在阿里电商业务取得了巨大的成功，这受益于智能算法的强大和工程能力的提升。本次分享将侧重在阿里个性化推荐平台（TPP）的工程层面，向大家介绍推荐工程平台如何灵活支持业务的快速发展，如何同时支持大量推荐场景在线优化和测试，又如何在双11大促中顶住流量洪峰。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>推荐工程全景剖析；<\/li>\n\t<li>平台成长采坑实战；<\/li>\n\t<li>灵活便利工具体系。<\/li>\n<\/ul>\n\n<h2>听众收益<\/h2>\n\n<ul>\n\t<li>了解推荐业务工程平台架构上的演进；<\/li>\n\t<li>了解大规模分布式系统构建实践经验；<\/li>\n\t<li>了解推荐在业务领域获得的一些经验。<\/li>\n<\/ul>\n","guestName":"李天民","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5954a31836af7.png","guestIntroduce":"0","beginTime":"1499416500","endTime":"1499419200","address":"马德里5号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0"},"createdAt":"1492602961","updatedAt":"1492602961","status":"0","days":"07","spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0","guestCompany":"阿里巴巴","guestPosition":"技术总监","dayorder":1,"ppt_link":""}],"producter_name":"李晗","topic_id":"90","order":"25","topic_description":"<p>如何获取用户操作习惯，如何采取合适推荐算法，如何大规模推送等均是推荐系统架构中需要研究的问题。<\/p>\n\n<p>广告则可以理解为内容的特别形式，在推荐系统架构中还需研究如何优化广告引擎与算法，兼顾用户体验和利润最大化。<\/p>\n\n<p>我们将邀请在此领域深耕多年的技术专家，揭开内容为王时代下精准推送背后的技术架构。<\/p>\n","introduce":"基于用户操作行为来预测用户可能喜欢的内容，在不少互联网产品中均有成熟的应用，但如何深刻了解用户的操作习惯，如何采取合适的推荐算法，如何大规模推送等均是推荐系统架构中需要研究的问题，本专题从经典案例入手带你一同解惑。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58d4829dec984.png","company":"阿里巴巴","position":"高级算法专家"},{"topic":"容器技术实践","content":[{"id":"910","conid":"18","topic":"容器技术实践","title":"微博机器学习平台混合云应用实践","introduce":"<p>新浪微博基于混合云搭建机器学习平台，用于服务线上信息流、关系流和消息推送流，目前支撑近百亿样本的特征工程。<\/p>\n\n<p>超大维度的特征工程，需要按需快速提供大规模的计算能力和存储资源，在已有一定规模的互联网企业，这样的需求对有限的机架资源也提出了很大挑战。<\/p>\n\n<p>微博利用 DCP 混合云调度平台，使用高配置的服务器，高密度的构建大规模的机器学习的支撑架构，满足了机器学习平台的业务要求。<\/p>\n\n<p>本次分享介绍微博在使用混合云支持机器学习平台的过程中遇到的技术、业务挑战，探讨对应的解决经验。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>微博机器学习平台的业务特点、规模；<\/li>\n\t<li>微博机器学习平台建设的挑战；<\/li>\n\t<li>如何柔性提供计算能力；<\/li>\n\t<li>如何应对机架资源不足；<\/li>\n\t<li>如何提供支撑混合云的网络架构；<\/li>\n\t<li>DCP 混合云调度平台在机器学习中的应用实践。<\/li>\n<\/ul>\n\n<h2>观众受益<\/h2>\n\n<ul>\n\t<li>了解超大维度特征工程的基础架构实践；<\/li>\n\t<li>了解混合云调度平台实践和相关业务收益；<\/li>\n\t<li>了解机器学习模型化的相关实践。<\/li>\n<\/ul>\n","guestName":"韩冰","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58fdbffd5a8df.png","guestIntroduce":"0","beginTime":"1499405400","endTime":"1499408100","address":"醉翁亭","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1492132814","updatedAt":"1492132814","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"新浪微博","guestPosition":"资深产品运维工程师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/983\/5965ef95cca8c.pdf\/18"},{"id":"839","conid":"18","topic":"容器技术实践","title":"京东容器平台与数据中心基础设施协同发展实践","introduce":"<p>随着容器技术的不断成熟，很多公司都在做容器，但总是不能非常彻底，很大一个障碍在于底层数据中心固有的老架构，京东在容器化过程也遇到这个问题并决心将这个痛点解决掉。<\/p>\n\n<p>京东商城业务规模庞大，架构复杂，特别是 JDOS（JD Data Center OS）系统的大规模落地使用，帮助商城研发显著提升系统上线效率和3倍提升计算资源使用率，对传统数据中心能力提出新的挑战。<\/p>\n\n<h2>问题挑战<\/h2>\n\n<ul>\n\t<li>传统的数据中心无法应对如此敏捷的容器化场景；<\/li>\n\t<li>传统的 DNS 服务，无法适应容器化业务敏捷服务发现功能；<\/li>\n\t<li>耦合的 L4-L7 负载均衡，无法满足业务快速弹性伸缩生效；<\/li>\n\t<li>网络环境传统二层架构、手工配置等，无法支撑业务快速定位网络延迟和百万级虚拟 IP 等问题。<\/li>\n<\/ul>\n\n<h2>演讲提纲（本次将主要分析2个创新技术点）<\/h2>\n\n<ul>\n\t<li>适应云架构的分布式智能 DNS 服务，高性能负载均衡服务等；<\/li>\n\t<li>打造全新计算平台整体解决方案，深化数据中心智能敏捷价值。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>分享京东在数据中心建设方面的思考、创新和实践；<\/li>\n\t<li>数据中心如何更好支持云架构系统的运行和发展。<\/li>\n<\/ul>\n","guestName":"鲍永成","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58c5f98413f7e.png","guestIntroduce":"0","beginTime":"1499409000","endTime":"1499411700","address":"醉翁亭","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58dd386c789ee.png","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1490861235","updatedAt":"1490861235","status":"0","days":"07","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58dd386c789ee.png","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"京东商城","guestPosition":"基础平台部技术总监","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/979\/596380422048e.pdf\/18"},{"id":"826","conid":"18","topic":"容器技术实践","title":"Apache Helix 对于在线\/离线生态系统的决定性作用","introduce":"<p>日益渐增的网络访问以及需求使得越来越多的应用需要考虑到自身的对于大数据，大工作量的延展性，高可用性会成为现如今各类应用在社会中的核心竞争力。<\/p>\n\n<p>在本次分享中将展现大数据在线\/离线生态系统中 Apache Helix 的作用以及优势，最后会展示现有均衡算法在各类著名系统中的应用，比如 LinkedIn 取代 Apache Voldermort 的新一代 key-value store，Venice是在线系统的代表。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>Helix 对大数据在线系统的应用；<\/li>\n\t<li>Helix 对大数据离线系统的应用；<\/li>\n\t<li>Helix 均衡算法在现有系统中的帮助<\/li>\n<\/ul>\n","guestName":"薛君凯","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7663d3c5ed.png","guestIntroduce":"0","beginTime":"1499412900","endTime":"1499415600","address":"醉翁亭","link":"","extension":{"spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"1"},"createdAt":"1490590024","updatedAt":"1490590024","status":"0","days":"07","spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"1","guestCompany":"LinkedIn","guestPosition":"Software Engineer\/ Apache Helix Commiter","dayorder":1,"ppt_link":""},{"id":"1042","conid":"18","topic":"容器技术实践","title":"小米生态云的应用引擎实践","introduce":"<p>小米生态云是面向小米生态链及合作伙伴企业的一站式综合云服务平台，保护小米生态的线上业务和用户数据的安全，汇聚和挖掘数据的价值，并逐步发挥人工智能的力量。<\/p>\n\n<p>小米应用引擎是小米生态云的重要组成部分和核心功能之一，在小米生态云上提供应用部署，监控报警，日志采集分析，域名管理备案及SSL证书, 自动化安全扫描，并集成各种常用的存储和计算服务以及深度学习框架。​<\/p>\n\n<p>小米应用引擎已经经历了两代的发展，此次和大家分享小米应用引擎的设计思路和引擎架构的演进过程，期间遇到的问题和解决办法，以及未来的一些计划。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>小米生态云简介<\/li>\n\t<li>小米生态云应用引擎演进\n\t<p style=\"margin-left:2px\">a. 第一代应用引擎<\/p>\n\n\t<p style=\"margin-left:2px\">b. 基于k8s的新版应用引擎<\/p>\n\t<\/li>\n\t<li>展望\n\t<p style=\"margin-left:2px\">a. 应用引擎<\/p>\n\n\t<p style=\"margin-left:2px\">b. 人工智能​<\/p>\n\t<\/li>\n<\/ul>\n","guestName":"李波","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/59351b94b1fbe.png","guestIntroduce":"0","beginTime":"1499416500","endTime":"1499419200","address":"醉翁亭","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1496653261","updatedAt":"1496653261","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"小米","guestPosition":"生态云团队资深研发工程师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/941\/595f4f45395a1.pdf\/18"}],"producter_name":"鲍永成","topic_id":"93","order":"28","topic_description":"<p>云经过了多年酝酿和发展，已不再单纯提供计算能力，而是成为包含基础设施、运算平台乃至整套管理、软件解决方案的庞大体系。在这样的背景下，云架构有哪些新发展，各云厂商如何应对未来云计算的全面战场？<\/p>\n\n<p>让我们一起聚在 Archsummit 大会共同探讨 Cloud Native 应用架构，Docker 容器技术，应用集群编排等对不断发展中的云架构带来的挑战和机会，探讨云架构如何与数据中心基础设施有机结合才能更好的服务 Cloud Native。<\/p>\n","introduce":"让我们共同探讨 Cloud Native 应用架构，Docker 容器技术，应用集群编排等对不断发展中的云架构带来的挑战和机会，探讨云架构如何与数据中心基础设施有机结合才能更好的服务 Cloud Native。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58c5f98413f7e.png","company":"京东商城","position":"基础平台部技术总监"},{"topic":"大规模分布式系统","content":[{"id":"971","conid":"18","topic":"大规模分布式系统","title":"Dynamic Multi-Raft：强一致高可用系统设计与实践","introduce":"<p>极致的可用性是所有分布式存储系统追求，业界从传统的主从复制，到现代的 Multi-Paxos \/ Raft 等基于分布式选举算法的日志复制状态机算法，而单一的 Raft 或者 Muti-Paxos 这些算法并没有描述如何在一个可弹性伸缩的存储系统上实现，这在全球范围内都是工业界和学术界的一个前沿的研究领域。<\/p>\n\n<p>在 TiKV 这个依赖 Multi-Raft 的新一代的 KV 系统中，如何处理多 Raft 组间的动态分裂和合并，做到可用性，扩展性和高性能的平衡，在这些方面我们积累很多第一手的宝贵经验，本次演讲会分享给大家。<\/p>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解分布式系统设计，高可用性系统设计，一致性算法<\/li>\n<\/ul>\n","guestName":"黄东旭","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/59190901ae32d.png","guestIntroduce":"0","beginTime":"1499405400","endTime":"1499408100","address":"巴塞罗那厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1494813330","updatedAt":"1494813330","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":" PingCAP","guestPosition":"创始人&CTO","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/922\/595f254419996.pdf\/18"},{"id":"969","conid":"18","topic":"大规模分布式系统","title":"分布式实现那些事儿：Pegasus 背后的故事","introduce":"<p>Pegasus 是一个小米云存储团队自主研发的分布式 K-V 存储系统，为小米内部的离线和在线业务提供了高可用、高性能、强一致的服务。我们采用了经典的中心化式的设计架构，并结合互联网业务的数据特点，使用了基于 hash 的 key schema；在实现语言上，我们使用 C++ 来做性能上的保障。现在 Pegasus 已经在公司内部对广告、视频等业务做了稳定的支撑，代码也已经全部开源。<\/p>\n\n<p>本次分享从 Pegasus 实现的角度的出发，分享了我们在构建强一致的分布式存储系统上的经验总结。<\/p>\n\n<p>演讲会首先介绍 Pegasus 的整体架构，以及我们在选型上的思考总结。然后我们还会介绍 Pegasus 背后的分布式框架 rDSN: 包括它的 IO 模型、RPC 框架，以及这些组件在分布式系统支持上的一些特点。我们也会着重介绍我们的 Deterministic 测试技术，以及它对我们一致性协议正确性上的帮助。另外，我们还会分享自己在项目开发上一些经验教训的总结。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>Pegasus 简单介绍；<\/li>\n\t<li>rDSN 的分布式组件；<\/li>\n\t<li>Deterministic 测试；<\/li>\n\t<li>开发上的经验教训。<\/li>\n<\/ul>\n\n<h2>听众收益<\/h2>\n\n<ul>\n\t<li>分布式存储系统架构设计的经验教训；<\/li>\n\t<li>分布式系统在实现、测试上的总结；<\/li>\n\t<li>了解小米云平台的技术动向。<\/li>\n<\/ul>\n","guestName":"孙伟杰","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/591280a3beb4c.png","guestIntroduce":"0","beginTime":"1499409000","endTime":"1499411700","address":"巴塞罗那厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1494384973","updatedAt":"1494384973","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"小米","guestPosition":"云存储工程师","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/929\/595f322fa7bc0.pdf\/18"},{"id":"825","conid":"18","topic":"大规模分布式系统","title":"存储世界，不止如此 | 腾讯 EB 级存储引擎背后的技术","introduce":"<p>本次演讲将主要分享从2006开始，腾讯内部从无存储平台到存储量达 EB 级别的 TFS2.0 存储平台这一过程中所经历的技术问题。<\/p>\n\n<p>在社交图片和视频盛行的时代，存储系统的设计和运营如何进行适配，揭秘微信 C2C 图片和视频如何提升体验、降低成本，以及在云时代腾讯是如何开放内部的存储技术的。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>EB 级存储系统的发展过程；<\/li>\n\t<li>图片时代存储设计以及在运营上的思考；<\/li>\n\t<li>视频时代的存储设计以及在运营上的思考；<\/li>\n\t<li>云时代的开放之路<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解 EB 级存储系统的发展过程，架构设计和困难挑战；<\/li>\n\t<li>了解图片以及视频时代，用户产生数据的特点和对应架构设计的思考；<\/li>\n\t<li>云时代的开发之路，如何从向用户服务升级到向企业服务。<\/li>\n<\/ul>\n","guestName":"邹方明","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f764a107437.png","guestIntroduce":"0","beginTime":"1499412900","endTime":"1499415600","address":"巴塞罗那厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1490589515","updatedAt":"1490589515","status":"0","days":"07","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"腾讯","guestPosition":"存储业务中心总监","dayorder":1,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/937\/595f45403a7c5.pdf\/18"},{"id":"1097","conid":"18","topic":"大规模分布式系统","title":"实时强一致性：电子客票系统中的挑战 ","introduce":"<p>电子客票系统作为特殊的服务与传统电商最大不同之处在于，它对数据一致性、可靠性及实时响应要求很高。本次分享将从国际机票入手介绍电子客票系统在数据对接、搜索、预定到下单等全链路过程中，如何在数据实时强一致性和可用性的权衡中取得平衡； &nbsp;&nbsp;&nbsp;<\/p>\n\n<h2>听众收益<\/h2>\n\n<ul>\n\t<li>电子客票系统设计的经验和教训；<\/li>\n<\/ul>\n","guestName":"杨进","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5959d43c31b12.png","guestIntroduce":"0","beginTime":"1499416500","endTime":"0","address":"巴塞罗那厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0"},"createdAt":"1499059378","updatedAt":"1499059378","status":"0","days":"07","spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0","guestCompany":"阿里巴巴","guestPosition":"飞猪平台技术团队技术总监 ","dayorder":1,"ppt_link":""}],"producter_name":"颜世光","topic_id":"94","order":"23","topic_description":"<p>无论是计算系统、存储系统，都离不开分布式架构设计。<\/p>\n\n<p>分布式系统是现代应用的标准形式，也是互联网架构的设计核心难点，复杂企业业务和海量数据等因素会让架构设计更加困难。<\/p>\n\n<p>我们将探讨互联网应用中的分布式架构设计理念，看各位技术专家在设计大规模分布式系统上有何心得。<\/p>\n","introduce":"分布式系统是现代应用的标准形式，也是互联网架构的核心难点之一，无论是计算系统、存储系统，都离不开分布式架构设计。本专题我们将探讨互联网应用中的分布式架构设计理念。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58bf6b04c164b.png","company":"百度","position":"搜索架构师"}],"day2":[{"topic":"机器学习与商业智能","content":[{"id":"1040","conid":"18","topic":"机器学习与商业智能","title":"AI筑巢：机器学习在百度凤巢的深度应用","introduce":"<p>凤巢作为国内最大的搜索广告平台，机器学习在其中扮演什么样的角色呢？百度从09年开始就将机器学习应用到广告CTR模型中来，到目前为止，商业广告的触发、点击预估、内容生态建设等几乎每个环节都深度使用机器学习算法。<\/p>\n\n<p>对于凤巢来讲，作为业界前沿的广告CTR预估模型，深度学习是如何首次全面作用于预估全流程？传统触发如何从基于语义的相关性到生成式触发模型？内容生态中标题、描述的智能撰写，图像算法覆盖图片素材生成、甄选各环节如何实现？本次演讲将带来机器学习在凤巢全方位的使用情况介绍，分享百度如何聚焦AI打造顶尖的商业智能系统。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>搜索广告业务简介\n\t<p style=\"margin-left:2px\">业务场景介绍<\/p>\n\t<\/li>\n\t<li>凤巢CTR预估引领机器学习在商业系统的应用\n\t<p style=\"margin-left:2px\">a.凤巢CTR模型算法演进历程<\/p>\n\n\t<p style=\"margin-left:2px\">b.排序全流程实现深度学习&nbsp;<\/p>\n\n\t<p style=\"margin-left:2px\">c.自有大规模机器学习平台的建设&nbsp;<\/p>\n\n\t<p style=\"margin-left:2px\">d.模型的规模与时效性（增量模型与实时学习）<\/p>\n\t<\/li>\n\t<li>凤巢触发遇见深度学习，华丽转身在前沿\n\t<p style=\"margin-left:2px\">a.凤巢传统触发演进历程<\/p>\n\n\t<p style=\"margin-left:2px\">b.深度学习带来生成式触发<\/p>\n\n\t<p style=\"margin-left:2px\">c.基于机器学习的意图触发<\/p>\n\t<\/li>\n\t<li>机器学习在商业内容生态的创新\n\t<p style=\"margin-left:2px\">a.标题创意的AI撰写<\/p>\n\n\t<p style=\"margin-left:2px\">b.凤巢图片-图像机器学习全覆盖<\/p>\n\n\t<p style=\"margin-left:2px\">c.机器学习打造智能落地页<\/p>\n\t<\/li>\n\t<li>未来的探索\n\t<p style=\"margin-left:2px\">a.图片、视频富媒体广告的机器合成技术探索<\/p>\n\n\t<p style=\"margin-left:2px\">b.视频AR、VR场景广告技术的探索<\/p>\n\t<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解机器学习等前沿技术如何赋能商业广告系统；<\/li>\n\t<li>了解商业广告的一些前沿探索和发展方向。<\/li>\n<\/ul>\n","guestName":"刘斌新","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5934d522c0c37.png","guestIntroduce":"0","beginTime":"1499477400","endTime":"1499480100","address":"大宴会厅2","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1496635089","updatedAt":"1496635089","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"百度","guestPosition":"商务搜索智能交互部负责人","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/947\/59603b5573cd5.pdf\/18"},{"id":"886","conid":"18","topic":"机器学习与商业智能","title":"数据驱动电商的零售运营","introduce":"<p>在电商和大数据的时代，随着电商平台（如 eBay、淘宝等）的出现和成熟，海量、实时、跨行业和零售商的市场数据比以前更容易获得，传统的零售运营也转向更为科学和系统的数据化运营，通过市场数据和信息帮助商家优化零售运营是电商平台必须为商家提供的服务之一。<\/p>\n\n<p>本次分享将着重介绍如何通过数据挖掘和机器学习帮助商家和运营团队来组织和优化促销活动。<\/p>\n\n<p>首先运用时间序列 Auto Regression 模型对历年的交易数据进行挖掘，推荐季节性促销的最佳时间段，并通过进一步的对用户搜索关键字基于 NLP 的挖掘，推荐最优化的时间段进行促销的商品类别。<\/p>\n\n<p>当运营团队选择了促销商品之后，我们通过对商品以往的成交行为特征，对促销活动的表现进行预测并推荐优化策略。在促销进行期间，利用大数据平台对促销活动的效果进行实时的监控和分析，帮助运营团队做出及时的调整。依托于数据的精细化运营贯穿促销活动的整个生命周期，形成闭环。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>电商平台在电商运营中的角色；<\/li>\n\t<li>电商零售运营的环节；<\/li>\n\t<li>数据驱动运用的实例：促销；\n\t<p style=\"margin-left:2px\">a. 挖掘季节性促销的最佳时间；<\/p>\n\n\t<p style=\"margin-left:2px\">b. 通过用户需求选择促销商品；<\/p>\n\n\t<p style=\"margin-left:2px\">c. 预测促销的表现；<\/p>\n\n\t<p style=\"margin-left:2px\">d. 实时监控和优化<\/p>\n\t<\/li>\n<\/ul>\n","guestName":"沈则潜","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f79ad212860.png","guestIntroduce":"0","beginTime":"1499481000","endTime":"1499483700","address":"大宴会厅2","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1491802956","updatedAt":"1491802956","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"eBay","guestPosition":"数据和商业智能研发总监","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/960\/59606f04cca6a.pdf\/18"},{"id":"1075","conid":"18","topic":"机器学习与商业智能","title":"海量社交网络中的热点话题发现","introduce":"<p>QQ空间是中国最大的社交网络，超过6亿用户在空间上分享动态。在海量的社交网络中，信息纷繁芜杂，流转多变，如何在海量的社交网络中快速找到用户最真切关心、热烈讨论的时下话题，是极具挑战的工作。通过热点话题，我们的产品、运营能够恰时地推出针对性的活动和服务，从而更好地满足用户的需求。<\/p>\n\n<p>本次演讲将介绍我们在海量社交网络中快速发现热点话题的技术方案和实践情况，分享我们对于海量社交文本的处理心得。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>热点话题业务应用场景介绍<\/li>\n\t<li>热点发现的技术方案介绍\n\t<p style=\"margin-left:2px\">a.海量社交短文本预处理<\/p>\n\n\t<p style=\"margin-left:2px\">b.社交新词发现<\/p>\n\n\t<p style=\"margin-left:2px\">c.深度学习话题提取<\/p>\n\n\t<p style=\"margin-left:2px\">d.热度综合打分重排序<\/p>\n\t<\/li>\n\t<li>未来的规划与探索\n\t<p style=\"margin-left:2px\">a.多模态内容融合<\/p>\n\n\t<p style=\"margin-left:2px\">b.实时流式话题发现<\/p>\n\t<\/li>\n<\/ul>\n\n<h2 style=\"margin-left:2px\">听众收益<\/h2>\n\n<ul>\n\t<li>了解海量社交网络中的热点发现技术方案<\/li>\n\t<li>了解深度学习在社交网络中的探索和应用<\/li>\n<\/ul>\n","guestName":"钟黎","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5949dac71a7ff.png","guestIntroduce":"0","beginTime":"1499484600","endTime":"1499487300","address":"大宴会厅2","link":"","extension":{"spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0"},"createdAt":"1498013422","updatedAt":"1498013422","status":"0","days":"08","spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0","guestCompany":"腾讯","guestPosition":"QQ数据中心NLP方向负责人","dayorder":2,"ppt_link":""}],"producter_name":"赵宇辰","topic_id":"88","order":"32","topic_description":"<p>随着机器学习技术的日新月异，人工智能已逐渐走入各个领域和被应用到不同的商业场景中，很多企业也将机器学习作为下一个发展的重点。<\/p>\n\n<p>同时机器学习的应用也有不同的侧重点：比如降低人力成本、提供更高效且实时的 insights、商业流程优化和自动化、个性化推荐等等。<\/p>\n\n<p>所有的这些智能应用都给企业商业带来间接或直接的优势，随着大数据时代的到来，如何打造一个稳定且可扩展的机器学习架构带来了很多新的挑战，因为机器学习的应用不仅仅是算法，而是一个复杂的系统工程。<\/p>\n\n<p>在本专题里，我们邀请了国内外机器学习和商业智能的专家，和大家一起分享大数据场景下的机器学习实战经验。<\/p>\n\n<p>&nbsp;<\/p>\n","introduce":"随着大数据时代的到来，如何打造一个稳定且可扩展的机器学习架构带来了很多新的挑战，在本专题里我们邀请了国内外机器学习和商业智能的专家和大家一起分享大数据场景下的机器学习实战经验。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58b78b8e8eeea.png","company":"销售易","position":"技术VP"},{"topic":"低延迟系统架构设计","content":[{"id":"888","conid":"18","topic":"低延迟系统架构设计","title":"蘑菇街分布式消息中间件 Corgi 在多场景下应用架构演进","introduce":"<p>蘑菇街早期架构基于 PHP 开发，从导购时代的 Redis，到电商初期的 RabbitMQ，到后期的自研，在电商快速发展的过程中，消息队列如何去适应不同场景的需求（如：多消费方式、低延迟、海量队列、MySQL Binlog 驱动业务等等）？<\/p>\n\n<p>我们的分布式消息中间件产品 Corgi 是如何选型、设计和快速迭代并满足复杂业务场景需求？<\/p>\n\n<p>本次分享将为大家带来详细解读。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>PHP 架构的电商系统在实时队列使用的演进，从 Redis 到 RabbitMQ 到自主研发；<\/li>\n\t<li>系统延迟与可用性之间的权衡以及多样化选择；<\/li>\n\t<li>基于 Dynamic Pull\/Push 满足不同消费能力 Consumer 的实时性；<\/li>\n\t<li>海量队列满足海量对象独享单独队列；<\/li>\n\t<li>Paxos 协议在消息队列多副本中的应用；<\/li>\n\t<li>集群混合部署平衡机器利用率与数据可用性。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>掌握低延迟的消息系统设计要点；<\/li>\n\t<li>消息读取的多样化以及对应的延迟；<\/li>\n\t<li>Paxos 协议在消息中间件中的运用。<\/li>\n<\/ul>\n","guestName":"江龙","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f79bd25e533.png","guestIntroduce":"0","beginTime":"1499477400","endTime":"1499480100","address":"马德里5号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1491872104","updatedAt":"1491872104","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"美丽联合集团","guestPosition":"技术专家","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/958\/59606cac1a386.pdf\/18"},{"id":"802","conid":"18","topic":"低延迟系统架构设计","title":"海量地址的狂欢：京东智能分单平台性能提升之路","introduce":"<p>本次分享旨在揭秘如何基于海量数据打造低延迟、高可用、高精准度的智能分单系统。中国物流规模已居世界前列，海量的数据与复杂性给电商系统提出了较高的性能要求。<\/p>\n\n<p>如在下单环节，用户填写的地址参差不齐，如何快速有效地识别正确地址，给行政区划错误、地址层级缺失、小区名称错别字以及不同城市道路河流差异性的地址做归类是一大痛点。<\/p>\n\n<p>京东智能分单系统应运而生，根据用户下单地址计算配送信息的系统，在用户下单时可以通过系统计算出仓储到配送员的全链路信息，迅速计算出包裹需要&ldquo;飞走&rdquo;的最佳路径。<\/p>\n\n<p>而面对这样复杂的业务系统，在保障业务稳定性的同时更要保证服务的响应速度，则低延迟的架构设计是始终不变的追求。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>订单生产流程架构与性能重要指标；<\/li>\n\t<li>应对措施：服务拆分规则化保障快速响应的架构方案；<\/li>\n\t<li>应对措施：缓存及一致性；<\/li>\n\t<li>应对措施：性能瓶颈之死锁及JVM调优实践；<\/li>\n\t<li>回顾与总结。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>构建低延迟高并发的系统；<\/li>\n\t<li>基于海量数据、打造低延迟的架构介绍；<\/li>\n\t<li>低延迟系统之性能毛刺实践。<\/li>\n<\/ul>\n","guestName":"王梓晨","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/59213da0d6de1.png","guestIntroduce":"0","beginTime":"1499481000","endTime":"1499483700","address":"马德里5号厅","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d2574a1a4f0.png","ppt_status":"1","video_status":"0","overseas":"0"},"createdAt":"1489728587","updatedAt":"1489728587","status":"-1","days":"08","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d2574a1a4f0.png","ppt_status":"1","video_status":"0","overseas":"0","guestCompany":"京东商城","guestPosition":"运营研发部系统架构师","dayorder":2,"ppt_link":""},{"id":"940","conid":"18","topic":"低延迟系统架构设计","title":"红包狂欢背后的秘密 | 手Q春节红包后台架构实践","introduce":"<p>通过不断创新玩法，手Q春节活动红包已成为春节一大玩点。2017年春节，AR 红包、刷一刷红包再创新高，抢红包用户数达3.42亿，共刷出红包37.77亿个。<\/p>\n\n<p>那么，在这亿级数字的背后，红包后台架构是如何设计的？为了顶住每秒百万级洪峰，又是如何做到低延时、高并发和高可用的？期间遇到了哪些挑战、采取了哪些优化措施，最终保证了活动的顺利进行？<\/p>\n\n<p>本次为实战分享，将为大家一一揭秘。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>手Q春节活动红包发展历程及产品简介；<\/li>\n\t<li>手Q春节活动红包面临的主要挑战；<\/li>\n\t<li>手Q春节活动红包后台架构设计思考及优化；\n\t<p style=\"margin-left:2px\">a.实时高效灵活的红包发放机制<\/p>\n\n\t<p style=\"margin-left:2px\">b.低延时高性能的通用地图任务方案设计<\/p>\n\n\t<p style=\"margin-left:2px\">c.高并发低延时的余量采集方案设计<\/p>\n\t<\/li>\n\t<li>大型活动类系统通用架构设计原则总结。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解海量秒杀场景下低延时高可用高并发系统的特点及架构挑战；<\/li>\n\t<li>获得大型活动类系统设计的方法和经验。<\/li>\n<\/ul>\n","guestName":"周海发","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5902a75259892.png","guestIntroduce":"0","beginTime":"1499484600","endTime":"1499487300","address":"马德里5号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1493346666","updatedAt":"1493346666","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"腾讯","guestPosition":"SNG社交网络事业群高级工程师","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/956\/596056d867a2b.pdf\/18"}],"producter_name":"者文明","topic_id":"91","order":"37","topic_description":"<p>随着互联网技术的不断发展和进步，低延迟已经被越来越多的技术人所关注。<\/p>\n\n<p>对于前端系统，低延时意味着体验的提升；<\/p>\n\n<p>对于网站系统，低延迟意味着流量的提升；<\/p>\n\n<p>对于交易系统，低延迟意味着销售量的增加。<\/p>\n\n<p>延迟不仅仅与用户体验、流量、销售量等直接相关，也是衡量高可用架构的重要标准。可见，构建低延迟应用、快速响应请求、提升系统吞吐能力已经成为我们在开发和架构设计过程中关注的焦点和不断追求的目标。<\/p>\n\n<p>而低延时应用的架构设计会涉及到诸多因素，比如操作系统、程序语言、IO 模型、读写机制、缓存机制、锁机制、网络环境、硬件条件等等。大型互联网架构已经进入毫秒之争，在降低信息处理延迟、通信延迟以及传输延迟等方面，目前各互联网企业在架构设计上都有哪些新思路？<\/p>\n","introduce":"延迟不仅仅与用户体验直接相关，也是衡量高可用架构的重要标准。大型互联网架构已经进入毫秒之争，在减少信息处理延迟、通信延迟以及传输延迟等方面，目前各互联网企业在架构设计上有哪些新思路？","guestavatar":"http:\/\/static.geekbang.org\/ck\/58e735b055290.png","company":"京东商城","position":"运营研发部总架构师"},{"topic":"安全之战","content":[{"id":"921","conid":"18","topic":"安全之战","title":"巡风系统在同程运维的安全实践","introduce":"<h2>概要<\/h2>\n\n<p>运维最担心的是什么？安全人员最喜欢的是什么？<\/p>\n\n<p>运维担心系统挂掉、用户打不开网站、报警电话等等，但最担心不是这些，最担心的是数据泄露与数据丢失，这两条是用户和公司都不能容忍的。<\/p>\n\n<p>数据泄露意味着可能涉及用户的隐私的泄露及公司的商业机密泄露。每一次网上的安全事件，只要是涉及到用户的个人信息，必定是重大安全事件；而数据丢失，比如硬盘坏了、恶意删除等，如果重要数据比如用户的支付订单数据丢失，公司的业务玩不下去了，还要我们运维和安全干什么，也没有我们什么事了。<\/p>\n\n<p>对于运维而言，数据泄露也许可以挽救，但是数据丢失呢，都丢了还怎么挽回？有人说备份，如果有备份就不叫丢失了。这里数据丢失，从运维安全角度说，是最让人不可忍受的。<\/p>\n\n<p>自动化运维安全思考，运维在部署运维系统时，越来越多的使用自动化系统，运维机器人帮助运维完成了大部分日常的任务，像发布、扩容、故障恢复、漏洞扫描等等，建议自动化运维遵循三大安全原则：<\/p>\n\n<p>一是最小化原则，原则上自动化部署，任务操作同一命令不得同时执行机器超过100台，中控台一定要把这一原则做为核心原则；<\/p>\n\n<p>二是确定性原则，自动化运维并不代表机器代替了人，而是人控制了机器，人比机器更清楚机器在做什么。<\/p>\n\n<p>在数据中心几千台，几万实例的运维今天，任何一个小的安全漏洞都会被放大，如何在这样的一个规模的今天，运维在自动化运维，已经向数据驱动的智能运维时代挺进，我们安全如何做，运维每天都在调整，各种层出不穷的开源系统，开源系统的安全配置又不可能每一个安全人员很清楚的掌握清楚，这就需要社区力量。<\/p>\n\n<p>同程网推出的巡风系统，一方面是在同程的自已的安全建议过程中逐步发展起来的，了解在一定规模下运维之痛、开发之痛，来源于现实；另外一方面积极开源，拥抱开源，在开源社区中接收营养，在开源的世界中吸纳更多的安全插件来补充巡风的插件，更好地补充这块企业人力的不足，同时也回馈社区，让更多的企业个人从中受益。<\/p>\n\n<p>目前巡风系统支持常规如 mongo 未授权访问、心脏滴血、同时也包括最近的 NSA 泄露漏洞、SMB 远程溢出漏洞等上百个系统漏洞扫描插件，比较多的企业开始关注并支持巡风的发展。<\/p>\n\n<h2>收获<\/h2>\n\n<ul>\n\t<li>安全对于运维整个体系的影响是什么？<\/li>\n\t<li>如何做好一个兼顾安全的运维系统？<\/li>\n\t<li>运维在安全方面要做好哪几件事？<\/li>\n\t<li>开源的巡风扫描器可以做什么？<\/li>\n<\/ul>\n","guestName":"刘新高","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58fdbe2877b11.png","guestIntroduce":"0","beginTime":"1499477400","endTime":"1499480100","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58fdd1755ebdb.png","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1493024660","updatedAt":"1493024660","status":"-1","days":"08","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58fdd1755ebdb.png","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"同程旅游","guestPosition":"运维经理","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/949\/59603f8e84865.pdf\/18"},{"id":"909","conid":"18","topic":"安全之战","title":"风控场景的模型平台架构","introduce":"<p>风控场景的扫描策略，一般是采用规则 + 模型的方式。在 DT + AI 时代，模型作为一种机器学习的代言，在其中扮演越来越重要的角色。由于模型是对数据进行机器学习训练得到的，这就产生了几个问题：<\/p>\n\n<ul>\n\t<li>如何原始数据加工成为模型所需要的数据？<\/li>\n\t<li>如何简单方便地进行模型训练，并评估模型效果？<\/li>\n\t<li>如何把线下的模型推送上线，并保证线上线下逻辑一致？<\/li>\n\t<li>如何有效地监控模型运行的情况，并在模型衰退之前作出反应？<\/li>\n\t<li>如何做到快速模型的在线更新，实时更新？<\/li>\n<\/ul>\n\n<p>模型平台要如何架构，才能更好解决上述问题，是安全架构师们面临的实际问题。这也是我们在实践中遇到的问题和挑战，在本次分享中，我将会从阿里巴巴的实践来进行详细说明。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>业务安全态势及全链路安全防控；<\/li>\n\t<li>安全防控手段；<\/li>\n\t<li>模型平台的定位及价值；<\/li>\n\t<li>模型平台的架构；<\/li>\n\t<li>架构经验（包含数据处理过程）<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解常见安全风控引擎建设过程中会碰到的问题；<\/li>\n\t<li>了解模型平台的架构要点；<\/li>\n\t<li>了解风控数据处理过程的难点及解法。<\/li>\n<\/ul>\n","guestName":"胡四海","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7a3c08332d.png","guestIntroduce":"0","beginTime":"1499481000","endTime":"1499483700","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1492132281","updatedAt":"1492132281","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"阿里巴巴","guestPosition":"高级技术专家","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/952\/59604d929f24c.pdf\/18"},{"id":"937","conid":"18","topic":"安全之战","title":" 人工智能在 Web 安全中的实践 ","introduce":"<p>传统的基于攻击特征的安全防护，存在着特征库难于管理、较高的误报率和漏报率等诸多问题，安全防护效果难以进一步提高，而近年来出现的基于语法规则的检测虽然在一定程度上弥补了特征的不足，但是因为不理解业务，对于有着更高智能的攻击也显得捉襟见肘。<\/p>\n\n<p>我将从百度在安全领域的实践出发，与大家一起探讨通过人工智能、机器学习领域的经典方法形成的分析系统效果，和他们如何与传统防护系统相互补充的经验。&nbsp;<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>基于神经网络的入侵检测识别原理与实现；<\/li>\n\t<li>神经网络识别及与语法引擎的对比分析；<\/li>\n\t<li>用户行为分析在 Web 日志审计中的模型介绍 。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解几种常见的分析模型在 Web 防护领域的实践方案；<\/li>\n\t<li>深度神经网络能对传统防护方案起到哪些弥补；<\/li>\n\t<li>机器学习在解决 Web 安全问题时遇到的各种挑战。<\/li>\n<\/ul>\n","guestName":"冯景辉","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/590192f1a9e3b.png","guestIntroduce":"0","beginTime":"1499484600","endTime":"1499487300","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1493275602","updatedAt":"1493275602","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"百度","guestPosition":"安全事业部技术总监 ","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/955\/596056a354c4f.pdf\/18"}],"producter_name":"韩方","topic_id":"87","order":"35","topic_description":"<p>IBM 2017 X-Force 提到2016年泄露的数量达到了40亿条，上万软件漏洞被记录在案。<\/p>\n\n<p>在互联网不同领域，研发安全都是企业不容忽视的环节，面对网络威胁，企业应该部署怎样的安全防线？<\/p>\n\n<p>企业安全体系架构如何建设，有哪些罕见的攻击点和思维陷阱，优秀安全工程师有这些经验想和你分享。<\/p>\n","introduce":"在不同的互联网领域，安全都会是重要的话题。从业于互联网安全，必须同时了解攻和防，那么企业安全体系架构如何建设？有哪些罕见的攻击点和思维陷阱？本专题集结了各位优秀安全工程师的多年思考。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58b792305d050.png","company":"欢聚时代（YY直播）","position":"安全中心总监"},{"topic":"架构师成长路线","content":[{"id":"924","conid":"18","topic":"架构师成长路线","title":"从大厂到创业公司，架构师经历的三次转身","introduce":"<p>提起&ldquo;架构师&rdquo;这个职位，大家往往指的是基础技术架构师，认为只有谈 CAP、Paxos、BlockChain、NLP、Clustering 才算是技术，只有研究各类分布式、大数据框架才是架构，有的甚至认为架构师就是&ldquo;框架师&rdquo;。<\/p>\n\n<p>但是真的是这样的么？针对大部分的公司（90%以上）来说，更多的工作却是&ldquo;平淡无奇&rdquo;的业务和&ldquo;增删改查&rdquo;的代码。&nbsp;很多开发人员在打怪升级过程中往往会进入一个迷茫期，自己花大量精力学习来的&ldquo;分布式架构&rdquo;、&ldquo;PB 量级的数据挖掘&rdquo;、&ldquo;亿级网站处理能力&rdquo;无用武之地，老板不重视，总是感叹&ldquo;怀才不遇&rdquo;，整个人陷入一个发展瓶颈。<\/p>\n\n<p>本次 Topic 是想通过本人的经历现身说法，聊聊架构师这个职位应该如何定义？是不是只有纯技术架构才是终极之道？优秀的架构师应该具备什么样的素质？<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>架构是做什么的？<\/li>\n\t<li>从一次自身经历失败的技术导入，引出开发人员的一种心态：拿着锤子找钉子。然后说明第一个结论：架构要选择最合适的而不是最好的；<\/li>\n\t<li>从一次自己做&ldquo;消防队员&rdquo;的经历，说明架构师不能是高高在上的，而是要撸起袖子解决实际问题；<\/li>\n\t<li>从目前的公司现状，描述了当前开发人员普遍忽视业务设计的重要性。然后介绍 DDD 设计方法和微服务架构的在实际中的运用情况，重点说明如何在业务设计中锻炼自己的架构技能能力。<\/li>\n\t<li>架构师不仅仅要做技术，还需要很多软技能，如何培养这些技能？<\/li>\n\t<li>QA<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>业务开发是不是就很 Low？如何在业务开发中提升自己的技术能力？<\/li>\n\t<li>把 DDD 和微服务架构的整合介绍给大家，改变大家对业务设计的偏见。<\/li>\n\t<li>工程师做技术的目标是什么？<\/li>\n\t<li>5年左右的高级开发工程师如何规划自己未来的技术生涯？<\/li>\n<\/ul>\n","guestName":"曹祖鹏","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58fed7c993d4b.png","guestIntroduce":"0","beginTime":"1499477400","endTime":"1499480100","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58fed93148eae.png","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1493096773","updatedAt":"1493096773","status":"-1","days":"08","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58fed93148eae.png","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"千米网","guestPosition":"首席架构师","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/948\/59603c5136134.pdf\/18"},{"id":"831","conid":"18","topic":"架构师成长路线","title":"从程序员到架构师，从架构师到 CTO","introduce":"<p>程序员、架构师、CTO，这个可能是很多技术人员为自己规划的职业发展路径，然而这三个职位各有各自的侧重点和思考方式，其衔接并非自然而然就能发生，而是需要有本质的飞跃。<\/p>\n\n<p>此次将分享从业15年一路走来，依次经历的这三个阶段，给大家分享一下其中的心路历程，如何做到眼界和技能的螺旋式上升。<\/p>\n","guestName":"洪强宁","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7687939fc5.png","guestIntroduce":"0","beginTime":"1499481000","endTime":"1499483700","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58f8349e26178.png","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1490707975","updatedAt":"1490707975","status":"-1","days":"08","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58f8349e26178.png","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"爱因互动","guestPosition":"创始人&CTO，EGO会员","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/951\/59604551d0435.pdf\/18"},{"id":"917","conid":"18","topic":"架构师成长路线","title":"造物先造人：八年架构与心路历程","introduce":"<p>在阿里工作8年，从工程师到架构师，一路下来经历了数次的业务变迁，技术也不断的升级。从负责公司内部核心系统到云数据库服务产品，从面对内部同事到面对外部用户，自身角色从合作者也逐渐变成了乙方，这个转变过程中需要不断的磨合自己与技术架构的关系。<\/p>\n\n<p>&ldquo;成长&rdquo;其实就是经验的积累，通过这次演讲，能把自己的宝贵财富分享给大家。在本次演讲中，根据时间顺序，一步步分享总结不同阶段的三个方面：业务、心态、架构，请大家和我一次来回顾&ldquo;成长&rdquo;历程，也希望各位听众能有所启发。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>自我介绍（做过什么，目前正在做什么） ；<\/li>\n\t<li>从 KV 系统视角看阿里集团的技术架构，打开眼界，夯实基础，从菜鸟成长为架构师；<\/li>\n\t<li>云产品的初级设计思路，将内部产品去包装成外部云服； 沿用了前面的经验，信心膨胀，但并不成功；<\/li>\n\t<li>云产品的进阶架构设计，重新认识客户，放空自己的经验，从零了解业务；<\/li>\n\t<li>总结，技术架构为人服务，由人设计，架构师的心态决定了架构的方向；<\/li>\n\t<li>Q&amp;A<\/li>\n<\/ul>\n","guestName":"杨成虎（叶翔）","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7a2f7ca2f8.png","guestIntroduce":"0","beginTime":"1499484600","endTime":"1499487300","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1492583669","updatedAt":"1492583669","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"阿里云","guestPosition":"云数据库高级数据库专家","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/954\/5960566e663a1.pdf\/18"}],"producter_name":"梁宇鹏（一乐）","topic_id":"83","order":"34","topic_description":"<p>过去我们讨论过了30岁能不能编程，今天我们讨论公司可不可以辞退34岁以上的员工，每个人都在审时度势走自己的路，因为在行业大发展的潮流中，不进则退。<\/p>\n\n<p>你加入一个创业公司，多一些辛苦有更大的空间，你进入一个大公司，可以触摸高精尖的技术，你与同事合伙，可以做一份自己的事业。所有的这些选择，也许只是不同的发展阶段，却是练级打怪的必经之路。<\/p>\n\n<p>如何成长为一个合格的架构师？一个架构师的未来又在哪里？<\/p>\n\n<p>架构师有架构师的问题，架构师有架构师的回答，让我们一起思考和探索。<\/p>\n\n<p>&nbsp;<\/p>\n","introduce":"如何成长为一个合格的架构师？一个架构师的未来又在哪里？架构师有架构师的问题，架构师有架构师的回答，让我们一起思考和探索。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58b78c8f40aed.png","company":"环信","position":"首席架构师\/云通讯事业部总经理，EGO会员"},{"topic":"大规模企业级性能优化","content":[{"id":"962","conid":"18","topic":"大规模企业级性能优化","title":"Web 加速，协议先行","introduce":"<p>访问 Web 网站和应用时，经常会遇到各种各样的性能问题。比如网页加载慢、视频卡、网络出错等，其中一个关键的影响因素就是网络协议。<\/p>\n\n<p>本次演讲会系统化地介绍 TCP、UDP、HTTP1.1、HTTPS（包括最新的 TLS1.3协议）、SPDY、HTTP2 等协议存在的问题，以及如何在特定的场景下通过网络协议的优化实现访问速度的提升。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>Web 性能问题；<\/li>\n\t<li>Web 协议分析；<\/li>\n\t<li>访问速度优化。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解 Web 性能的影响因素；<\/li>\n\t<li>了解 Web 协议的原理，合理使用前端的页面优化策略；<\/li>\n\t<li>合理使用和优化 Web 协议，提升产品的访问速度。<\/li>\n<\/ul>\n","guestName":"罗成","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/590b00515db6e.png","guestIntroduce":"0","beginTime":"1499477400","endTime":"1499480100","address":"巴塞罗那厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1493893541","updatedAt":"1493893541","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"腾讯","guestPosition":"TEG 基础架构部 高级工程师","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/946\/59603a02e88af.pdf\/18"},{"id":"965","conid":"18","topic":"大规模企业级性能优化","title":"无线动态化框架 LuaView 性能优化","introduce":"<p>随着无线浪潮的逐步推进，无线端应用对动态化能力的需求，与日俱增，各种动态化方案层出不穷。在这个大的浪潮中，阿里聚划算开发出了自己的动态化框架－LuaView。目前 LuaView 已经开源，后续希望 LuaView 不仅仅服务于阿里内部，更能为整个无线互联网生态服务。<\/p>\n\n<p>LuaView 的产生初衷是为了提升移动端应用动态化能力，提供比 H5 更好的用户体验，同时做到 Android、iOS 业务代码共用，减少同一业务的人员重复投入。<\/p>\n\n<p>目前 LuaView 已经在聚划算大范围使用，先后经历了聚划算 3.8、6.6、9.9、11.11大促，已经服务过亿万用户。使用 LuaView 的客户端能够具备 H5 同样的动态化能力，做到 bug 随时修，业务随时上，且在硬件交互，动画等功能上提供比 H5 更好的体验。<\/p>\n\n<p>这里主要介绍 LuaView 的起源、整体架构、介绍我们在 LuaView 性能优化上所做的工作和努力。同时还会介绍 LuaView 的应用场景以及面临的问题、挑战等内容。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>无线动态化框架的产生与发展；<\/li>\n\t<li>聚划算无线动态化之路；<\/li>\n\t<li>LuaView 的整体架构和设计理念；<\/li>\n\t<li>LuaView 的发展与性能优化实践。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解无线动态化框架的前世今生；<\/li>\n\t<li>了解主流无线动态化框架；<\/li>\n\t<li>了解 LuaView 整体架构和设计思路；<\/li>\n\t<li>了解 LuaView 的发展历程、性能优化等实战经验。<\/li>\n<\/ul>\n","guestName":"陈松涛（野松）","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/590fe95b21ba1.png","guestIntroduce":"0","beginTime":"1499481000","endTime":"1499483700","address":"巴塞罗那厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1494215279","updatedAt":"1494215279","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"阿里巴巴","guestPosition":"天猫技术部资深无线开发工程师","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/950\/59604ae286498.pdf\/18"},{"id":"801","conid":"18","topic":"大规模企业级性能优化","title":"Mobile Performance at Scale","introduce":"<p>Everyday, various Facebook mobile apps are being used by over one billion devices around the world. In this talk, I will share with you how Facebook thinks about about mobile performance.<\/p>\n\n<p>We&#39;ll look at some of the challenges we face and how they become an engineering problem at scale. We will explain how we tackle them by treating performance as a first-class feature.<\/p>\n\n<p>Then we&#39;ll take a high-level look at some of our performance tools, and then do a deep-dive into tools for regression prevention.<\/p>\n\n<p>每天在世界各地，超过10亿的手机在运行着 Facebook 的多个移动应用。这个演讲将和您分享 Facebook 对移动应用性能这个工程问题的理念。<\/p>\n\n<p>我们将解释我们所遇到的一些挑战，向您展示移动应用性能是如何成为一个大型的工程问题的，以及我们如何将性能第一深入到工程的各个环节。<\/p>\n\n<p>之后我们会介绍一些性能工具，并着重深入介绍性能退化检测的工具。<\/p>\n","guestName":"黄力菲，倪安迪","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/593137400f3f2.png","guestIntroduce":"0","beginTime":"1499484600","endTime":"1499487300","address":"巴塞罗那厅","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d257fbee40b.png","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1489728262","updatedAt":"1489728262","status":"0","days":"08","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d257fbee40b.png","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"Facebook","guestPosition":"Engineering Manager，Software Engineer","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/957\/59605ba216d01.pdf\/18"}],"producter_name":"唐文","topic_id":"81","order":"33","topic_description":"<p>在现代市场上，企业是由运行他们商业价值的产品所定义，而产品的用户体验则直接决定了企业的收入和声誉。如今，互联网蕴藏着巨大的财富，但企业应用逻辑及应用所在的环境日益复杂，业务数据量也日趋庞大。<\/p>\n\n<p>随着智能手机、平板电脑与智能设备的广泛使用，以及云计算和虚拟化的逐渐普及，使得影响应用性能的因素变得越来越复杂，并且形势愈发严峻。<\/p>\n\n<p>所有几乎参与产品的人、事务和资源都会产生用户体验问题，而往往这些不同维度的体验问题会叠加在一起，并不断放大，例如产品逻辑、开发语言、代码性能、基础网络、服务器及云环境等。<\/p>\n\n<p>如何理解大规模企业级性能优化？是指移动、前端、后端、网络、系统、应用、硬件、产品逻辑等协同优化？还是指协议级别的改造实现大规模的成本节省？本专题将精选行业中的最佳实践来诠释这一概念。<\/p>\n\n<p>&nbsp;<\/p>\n","introduce":"如何理解大规模企业级性能优化？是指移动、前端、后端、网络、系统、应用、硬件、产品逻辑等协同优化，还是指协议级别的改造实现大规模的成本节省？本专题将精选行业中的最佳实践来诠释这一概念。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58b78a9a62c5d.png","company":"高升控股","position":"技术VP"},{"topic":"基于微服务的软件新架构","content":[{"id":"966","conid":"18","topic":"基于微服务的软件新架构","title":"Using sagas to maintain data consistency in a microservice architecture","introduce":"<p>The microservice architecture structures an application as a set of loosely coupled, collaborating services. Maintaining data consistency is challenging since each service has its own database to ensure loose coupling. To make matters worse, for a variety of reasons distributed transactions are not an option for modern applications.<\/p>\n\n<p>In this talk we describe an alternative transaction model known as a saga. You will learn about the benefits and drawbacks of using sagas. We describe how sagas are eventually consistent rather than ACID and what this means for developers. You will learn how to design and implement sagas in a Java application.<\/p>\n","guestName":"Chris Richardson","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/590febad971be.png","guestIntroduce":"0","beginTime":"1499477400","endTime":"1499480100","address":"大宴会厅1","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1494215710","updatedAt":"1494215710","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"","guestPosition":"《POJOs in Action》作者，微服务专家","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/980\/596527011e25e.pdf\/18"},{"id":"1005","conid":"18","topic":"基于微服务的软件新架构","title":"微服务架构模式","introduce":"<p>此次分享主要介绍近几年 bilibili 在微服务过程中的微服务架构经验，分为不同的纬度去解析在微服务落地中的细节实现。<\/p>\n\n<p>比如网关如何聚合服务，难点在哪里？<\/p>\n\n<p>在服务治理里面，服务发现应该用 zookeeper 还是其他方式？<\/p>\n\n<p>配置文件在服务治理中如何动态加载，开关如何启用？<\/p>\n\n<p>缓存作为必不可少的环节，需要重点关注哪些性能细节？<\/p>\n\n<p>业务要求比较严格的一致性，缓存、存储、队列如何保证一致性？<\/p>\n\n<p>可用性又如何在众多微服务之间协作过程中达成？<\/p>\n\n<p>带着这些问题，慢慢听我道来&hellip;<\/p>\n\n<h2>听众受益<\/h2>\n\n<p>了解微服务架构设计中的，各个重要环节，包括网关设计、服务治理、缓存设计、一致性设计、可用性设计等细节<\/p>\n","guestName":"毛剑","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5923daa955e48.png","guestIntroduce":"0","beginTime":"1499481000","endTime":"1499483700","address":"大宴会厅1","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1495522157","updatedAt":"1495522157","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"哔哩哔哩","guestPosition":"技术总监","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/953\/59604d6aebe11.pdf\/18"},{"id":"901","conid":"18","topic":"基于微服务的软件新架构","title":"从微服务到 Serverless 架构：享受纯粹的编程乐趣","introduce":"<p>同程旅游在近些年中技术平台升级了好几代，也进行了微服务化改造，在微服务实践过程中解决了原来许多架构上的问题（耦合、扩展、熔断、降级、调用治理、多语言等）。<\/p>\n\n<p>但这些问题被解决掉后我们进行了更进一步的升级，将 Serverless 想法引入架构中并开发上线了一个基于我们私有云的 Serverless 平台，现在整个同程的 Web 站点和部分变化快的轻型都这个 Serverless 之上开发。&nbsp;<\/p>\n\n<p>为什么我们在微服务之后还要做 Serverless 架构呢？主要 Serverless 带来的好处正好解决我们最新的痛点：业务快速变化中如何更快地开发。其实在这快速地开发中有很多是快不起来的（开发环境的问题，上线部署的问题，应用弹性设计问题，可运维性的问题等等）。<\/p>\n\n<p>总之在这过程中编程者考虑程序以外的事情太多，总结为一句话&ldquo;一条 SQL 到一个服务的距离到底有多远&rdquo;，很远，我们要做的是让距离变得更短一些。<\/p>\n\n<p>另外一点在于互联网流量的变化太大，时大时小，这让大量部署的服务器资源使用率不够高，我们做到的是让应用在被使用的时候才产生资源的占有，不跑不占资源，本次主要分享同程在 Serverless 架构实践过程中的一些坑与事。<\/p>\n\n<h2>演讲题纲<\/h2>\n\n<ul>\n\t<li>同程实践 Serverless 的背景；<\/li>\n\t<li>Serverless 与传统架构比较；<\/li>\n\t<li>同程的 Serverless 实现了什么；<\/li>\n\t<li>Serverless 架构在同程的一些实际使用那些业务场景分析；<\/li>\n\t<li>我们的下一步在 Serverless 上要做什么。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>Serverless 与传统架构比较；<\/li>\n\t<li>Serverless 在同程实践的技术细节点；<\/li>\n\t<li>Serverless 适用于什么样的场景。<\/li>\n<\/ul>\n","guestName":"王晓波","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7a08c6259f.png","guestIntroduce":"0","beginTime":"1499484600","endTime":"1499487300","address":"大宴会厅1","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1491940703","updatedAt":"1491940703","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"同程旅游","guestPosition":"首席架构师，EGO会员","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/959\/59606cc920b78.pdf\/18"}],"producter_name":"谢孟军","topic_id":"79","order":"31","topic_description":"<p>随着各行各业将微服务应用于架构设计，促发了更多各领域微服务的创新实践。<\/p>\n\n<p>微服务中，如何解决服务管理、应用管理、负载均衡、雪崩等一系列痛点问题？<\/p>\n\n<p>现在有哪些值得参考的微服务创新架构，能够给架构师们更多启发？<\/p>\n","introduce":"随着各行各业将微服务这一设计理念应用于架构设计中，促发了更多微服务的新概念和外延，在一背景下，我们精选了基于微服务的创新架构案例，希望能给架构师更多启发。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58ff41ff29eb4.png","company":"GopherChina","position":"创始人，EGO会员"},{"topic":"技术创业","content":[{"id":"832","conid":"18","topic":"技术创业","title":"创业维艰","introduce":"<p>创业不易，很多因素影响着结果，你可以总结出几十个上百个死掉的情况，但成功的原因却没几个。我们试着从自己的经历中找到一些观点，来总结这些年我们创业的过程。<\/p>\n\n<p>从2007年开始，我们陆续经历了三次创业过程，我会从不同的几个角度来和大家分享。<\/p>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>创业团队在人员组成和内部问题上的一些坑；<\/li>\n\t<li>对于创业方向的一些思考。<\/li>\n<\/ul>\n","guestName":"丁宁","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f795348fa7a.png","guestIntroduce":"0","beginTime":"1499491800","endTime":"1499494500","address":"马德里5号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0"},"createdAt":"1490767715","updatedAt":"1490767715","status":"0","days":"08","spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0","guestCompany":"荔枝FM","guestPosition":"CTO，EGO会员","dayorder":2,"ppt_link":""},{"id":"800","conid":"18","topic":"技术创业","title":"从0到1再到100，创业不同阶段的技术管理思考","introduce":"<p>有米科技从2010年创业开始，从几个人到100多人的技术团队，如何规划好公司技术架构、人才梯队、组织架构，适应不同阶段的需求？<\/p>\n\n<p>从寻求生存期的快速试错，到应对业务增长迅猛的挑战，到遇到瓶颈期的寻求突破，此次分享将探讨我们在有米科技创业6年的回顾和思考。<\/p>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解一个典型的互联网公司从创业到快速发展全过程的复盘及亲历者的想法与思考；<\/li>\n\t<li>少走弯路，更好地为创业的不同阶段做准备。<\/li>\n<\/ul>\n","guestName":"蔡锐涛","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f76137d18b0.png","guestIntroduce":"0","beginTime":"1499495400","endTime":"1499498100","address":"马德里5号厅","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d254bfed2d4.png","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1489723977","updatedAt":"1489723977","status":"0","days":"08","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d254bfed2d4.png","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"有米科技","guestPosition":"创始合伙人&CTO，EGO会员","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/966\/596086bd2f046.pdf\/18"},{"id":"760","conid":"18","topic":"技术创业","title":"独角兽炼成记","introduce":"<p>对于初创公司来说，什么是最重要的，创始团队最主要的职责是什么？<\/p>\n\n<p>什么决定了初创公司能否快速发展？成功不可复制，失败经验值得学习借鉴，初创团队怎么可以少走弯路，避免那些常见的坑？<\/p>\n\n<p>有关人、团队、组织架构、企业文化在公司快速发展的过程中是如何演进的，如何跟业务相辅相成，相互促进？<\/p>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解到一家在3年内成为独角兽公司的成长历程；<\/li>\n\t<li>创业和创始团队要关注的核心点；<\/li>\n\t<li>了解和避开创业过程中那些常见的坑；<\/li>\n\t<li>创业是个系统工程，从系统的角度思考和指导实践。<\/li>\n<\/ul>\n","guestName":"黄伟坚","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f760978c69a.png","guestIntroduce":"0","beginTime":"1499499300","endTime":"1499502000","address":"马德里5号厅","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d252263940f.png","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1489476110","updatedAt":"1489476110","status":"-1","days":"08","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d252263940f.png","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"Mobvista","guestPosition":"联合创始人，EGO会员","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/971\/5960955469249.pdf\/18"},{"id":"838","conid":"18","topic":"技术创业","title":"从码农到CTO再到CEO过程中的认知升级","introduce":"<p>在一个销售渠道驱动的公司里，推动将技术服务业务，走向技术驱动业务的过程，从中孵化出一个 2B 的创新项目，如何建立一个扁平开放高效的研发团队，在创业的初期阶段如何与地推销售团队形成相互辉映，想和大家简单分享一些经验与思考。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>从码农到CEO,认知升级的三个阶段<\/li>\n\t<li>创业路上打怪升级篇<\/li>\n\t<li>创业认知方法论<\/li>\n\t<li>创业的真相<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>对于技术管理者构建一个扁平开放高效的团队，或许有些帮助；<\/li>\n\t<li>如何拿捏在产品技术团队与地推销售团队要同时推进的平衡感，确保项目步调高度协同，或许对同类创业者有帮助；<\/li>\n\t<li>透过自身经验来谈谈创业的认知积累，也许会让一些人有些感同身受。<\/li>\n<\/ul>\n","guestName":"刘小明","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f795ff3dcd3.png","guestIntroduce":"0","beginTime":"1499502900","endTime":"1499505600","address":"马德里5号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1490861155","updatedAt":"1490861155","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"广州经传集团","guestPosition":"CTO，EGO会员","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/978\/5960a31e04899.pdf\/18"}],"producter_name":"邹光先","topic_id":"84","order":"48","topic_description":"<p>我们经历了团购、O2O、P2P、直播等创业风口大战，目前共享单车、人工智能的竞争正在进行。<\/p>\n\n<p>有的技术创业能够一步直接跃升成为独角兽，有的技术创业苟延残喘苦苦挣扎至今未能脱身，为什么会有这些区别？<\/p>\n\n<p>如今2017年是更好的时期还是更差的一年？技术人创业的那些成功点与失败处有哪些可以有所借鉴？<\/p>\n","introduce":"对技术人而言，2017是更好的时期还是更差的一年？本专题将分享技术人创业的那些成功点与失败处，希望对创业者有所借鉴。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58b78d674f279.png","company":"爱拍","position":"联合创始人&CTO，EGO会员"},{"topic":"创新的智能应用","content":[{"id":"899","conid":"18","topic":"创新的智能应用","title":"人工智能在消费金融领域的实践","introduce":"<p>介绍人工智能在买单侠的应用，包括客户营销、信用评分、反欺诈以及贷后生命周期管理中如额度策略、风险追踪以及差异化定价等场景。<\/p>\n\n<p>这次分享将着重介绍 AI 在风险决策中的应用，如何帮助显著降低秦苍科技的坏账损失。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>互联网时代消费金融业务的挑战<\/li>\n\t<li>人工智能在信用评分中的应用<\/li>\n\t<li>人工智能在反欺诈中的应用<\/li>\n\t<li>在消费金融其他应用领域的人工智能探索<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>消费金融的挑战、AI 在消费金融场景中的应用、AI 在秦苍科技的应用。<\/li>\n<\/ul>\n","guestName":"李炫熠","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f79f5986835.png","guestIntroduce":"0","beginTime":"1499491800","endTime":"1499494500","address":"大宴会厅2","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1491939615","updatedAt":"1491939615","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"上海秦苍信息科技","guestPosition":"联合创始人&CTO，EGO会员","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/981\/5965ef1955e77.pdf\/18"},{"id":"908","conid":"18","topic":"创新的智能应用","title":"如何设计开发智能分析产品？供应链领域的实践启示","introduce":"<p>供应链管理是大多数制造、零售等多个领域的核心枢纽环节，其中诸如需求预测、库存控制、物流调度优化等问题，都对于新技术的应用有强烈诉求。<\/p>\n\n<p>数据分析和人工智能算法是升级供应链的利器。本报告将结合工业界的实际问题，开阔思路，阐述如何巧妙挖掘抽象问题，归类实践已有方法，在产品中实施各种算法；并探讨产品解决方案过程中可能遇到的各种问题和经验方法，对其他领域的智能应用具有普适性和参考作用。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>供应链简介和升级面临的问题；<\/li>\n\t<li>数据和人工智能算法在供应链体系中的应用方法论和思路；<\/li>\n\t<li>实际案例介绍，在特定场景下，如何将算法和业务结合，并获得最佳实践；<\/li>\n\t<li>未来数据算法在供应链以及其他相关场景中的应用展望。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解企业应用的重要战场-供应链管理领域；<\/li>\n\t<li>和听众共同抽丝剥茧分析如何结合实际业务需求来设计算法，并将算法应用到产品中；为其他领域人工智能应用提供参考；<\/li>\n\t<li>开阔思路，探讨未来智能数据算法在工业界的可能应用场景。<\/li>\n<\/ul>\n","guestName":"陈玉君","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7a220003cf.png","guestIntroduce":"0","beginTime":"1499495400","endTime":"1499498100","address":"大宴会厅2","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1492081717","updatedAt":"1492081717","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"华为","guestPosition":"数据科学家","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/982\/5965ef675f56a.pdf\/18"},{"id":"1001","conid":"18","topic":"创新的智能应用","title":"自然语言理解技术与智能问答实践","introduce":"<p>智能问答系统是自然语言处理技术的典型应用场景。智能问答系统与专家系统的天然结合，让人工智能在生产生活的各个领域渗透到无所不在成为可能。在各种智能客服、智能助理、聊天机器人出现的今天，智能问答系统的架构、语义理解设计与算法、知识引擎的构建是这一领域的热点问题，也是决定商业成功的核心技术。<\/p>\n\n<p>本次演讲侧重于讲述智能问答中的自然语言的提问理解技术，从传统的搜索引擎的关键词查询与文本匹配，如何过渡进化到真正的语义理解，从而实现更具通用价值的智能问答产品，支持可扩展的商业应用。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>智能问答系统的应用和关键问题；<\/li>\n\t<li>自然语言问题理解与搜索引擎查询理解的对比；<\/li>\n\t<li>自然语言问题理解典型问题的解决算法；<\/li>\n\t<li>自然语言理解技术的智能问答产品应用实践。<\/li>\n<\/ul>\n\n<h2>听众收益<\/h2>\n\n<ul>\n\t<li>智能问答系统的应用场景与设计选择；<\/li>\n\t<li>语义理解的典型问题解决方案；<\/li>\n\t<li>自然语言理解在智能问答系统中的实战经验。<\/li>\n<\/ul>\n","guestName":"刘琳","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/591e6d12110dc.png","guestIntroduce":"0","beginTime":"1499499300","endTime":"1499502000","address":"大宴会厅2","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1495166736","updatedAt":"1495166736","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"Yahoo! Inc.","guestPosition":"Sr. Engineering Manager","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/970\/5960951dba920.pdf\/18"},{"id":"916","conid":"18","topic":"创新的智能应用","title":"电商中的智能应用","introduce":"<p>小红书作为中国乃至全球独一无二的社区电商，有着自己独特的使命和优势。我们的目标是让全世界的好生活触手可及，那么如何帮助用户在社区里发现好生活，又如何引导用户找到值得剁手的好产品呢？<\/p>\n\n<p>我们深入的讨论如何运用机器学习和人工智能，帮助我们不到10人的算法团队快速接近目标。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>小红书的产品与团队，机器学习在小红书的应用效果：<\/li>\n\t<li>深度理解小红书用户产生的内容：<\/li>\n\t<li>小红书的推荐与搜索：<\/li>\n\t<li>如何在初创公司合理使用人工智能<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>在电商场景中的人工智能应用实例<\/li>\n\t<li>深度学习在电商内容搜索，推荐中的应用前景<\/li>\n\t<li>如何在初创公司合理使用人工智能，实战 Tips<\/li>\n<\/ul>\n","guestName":"赵晓萌","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7a5fc0f92c.png","guestIntroduce":"0","beginTime":"1499502900","endTime":"1499505600","address":"大宴会厅2","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1492582949","updatedAt":"1492582949","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"小红书","guestPosition":"算法架构师","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/985\/5967950fde27c.pdf\/18"}],"producter_name":"苑维然","topic_id":"82","order":"42","topic_description":"<p>人工智能在这一年中经历了华丽的蜕变。<\/p>\n\n<p>这一年，我们看到了学术界和工业界正在积极地互相配合，在计算机视觉、语音识别、自然语言理解等方向取得了非常有价值的进展，也看到了它在机器人、穿戴设备、自动驾驶、智能家居等行业里让人心动不已的展示。<\/p>\n\n<p>如今，在交通、医疗、金融、供应链、娱乐、电商等垂直行业应用的层面，人工智能正在率先打破泡沫，在合适的场景下和企业架构深度集成，真正用来解决发展的痛点，产生实际的价值，在这里我们带给您几个垂直行业应用的案例。<\/p>\n","introduce":"人工智能在这一年中经历了华丽的蜕变。如今，在交通、医疗、金融、供应链、娱乐、电商等垂直行业应用的层面，人工智能正在率先打破泡沫，在合适的场景下和企业架构深度集成，真正用来解决发展的痛点，产生实际的价值，在这里我们带给您几个垂直行业应用的案例。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58b787d14fa69.png","company":"格灵深瞳","position":"首席架构师"},{"topic":"研发工具与工程效率","content":[{"id":"914","conid":"18","topic":"研发工具与工程效率","title":"华为Git实践：工作模式创新，及多中心分布式架构","introduce":"<ul>\n\t<li>为什么 GitHub、谷歌的 Git 平台这么快？他们的可扩展的 Git 平台是如何设计的？<\/li>\n\t<li>GitHub 的分布式协同模式和 Gerrit 的集中式协同模式哪一个更适合企业？Git 是如何适配华为的？<\/li>\n<\/ul>\n","guestName":"蒋鑫","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f7a52161023.png","guestIntroduce":"0","beginTime":"1499491800","endTime":"1499494500","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1492500524","updatedAt":"1492500524","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"华为","guestPosition":"技术专家","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/961\/5960772f5d705.pdf\/18"},{"id":"881","conid":"18","topic":"研发工具与工程效率","title":"Tools for Advanced Prototyping for Mobile Apps","introduce":"<p>Prototyping allows app development teams to test out improvements, and gather feedback with less development resources and at a much faster velocity.<\/p>\n\n<p>In a competitive app environment, the speed and cost to developing and validating valuable new improvements to your app can be the difference between success and failure.<\/p>\n\n<p>This talk will consider different tools and how teams can apply them prototypes.<\/p>\n\n<p>&nbsp;<\/p>\n\n<p>参考翻译：<\/p>\n\n<p>Prototype 原型化开发可以让 app 开发团队对改进情况进行彻底测试，并以更少的开发资源和更有效率的方式获得反馈。<\/p>\n\n<p>在一个竞争的 app 环境里，开发中的速度和成本，和在 app 里验证有价值的新环境，可以决定成功与失败。<\/p>\n\n<p>在这次分享里，我们将比较不同的工具以及团队如何在原型化开发迭代中使用他们。<\/p>\n","guestName":"Joe","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f799afa9b34.png","guestIntroduce":"0","beginTime":"1499495400","endTime":"1499498100","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58fef11b089be.png","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1491480963","updatedAt":"1491480963","status":"-1","days":"08","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58fef11b089be.png","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"","guestPosition":"Startup Veteran","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/964\/596083a8ec10f.pdf\/18"},{"id":"957","conid":"18","topic":"研发工具与工程效率","title":"互联网时代的持续交付","introduce":"<p>阿里持续交付平台已经经历了8年的不断迭代进化，成长为集团几万应用所依赖的最重要的研发工具，它的效率直接影响着几万研发日常工作。<\/p>\n\n<p>但平台不能只是工具的堆砌，更需要针对互联网时代的研发模式进行深度思考，不断打磨，将工程师文化和工程师实践不断地融入其中。<\/p>\n\n<p>轻管控重技术，使用业界上最新工程实践，用技术的演进去解决技术人的效率问题。本次演讲将介绍阿里持续交付工具的演化历程和对互联网行业交付领域热点问题的思考实践。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>持续交付在阿里；<\/li>\n\t<li>质量与效率；<\/li>\n\t<li>交付之痛；<\/li>\n\t<li>DevOps 机遇与挑战。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>阿里面向4亿用户服务的持续交付经验和平台建设思路；<\/li>\n\t<li>互联网时代软件质量和效率如何平衡；<\/li>\n\t<li>国际化和私有云对软件交付的挑战和应对策略；<\/li>\n\t<li>DevOps 实践中工具平台所起到的作用。<\/li>\n<\/ul>\n","guestName":"陈鑫（神秀）","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/59081d9821e7e.png","guestIntroduce":"0","beginTime":"1499499300","endTime":"1499502000","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1493704297","updatedAt":"1493704297","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"阿里巴巴","guestPosition":"高级技术专家","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/969\/5960952905226.pdf\/18"},{"id":"755","conid":"18","topic":"研发工具与工程效率","title":"Facebook 的代码开发工具","introduce":"<p>在 Facebook，我们 move fast，为此我们改造编程语言，创建新的编程语言和运行环境，因为市面上没有很好的开发工具能够支持我们自己的语言和大型代码，我们开发了自己的开发工具 Nuclide 和代码审核工具 Phabricator。<\/p>\n\n<p>这两个工具均已被开源，这次分享将重点介绍 Nuclide 的开发，相信能让您对 Facebook 的开发工具和开发模式有所了解，希望对您公司的开发有所启发。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>介绍 Nuclide 的历史，包括一些重要的设计决定是如何做出的；<\/li>\n\t<li>介绍开发流程上的一些创新；<\/li>\n\t<li>深入介绍 Nuclide 的架构；<\/li>\n\t<li>在解释这些的同时，我们不可避免的需要对其他一些工具做些介绍，比如 Phabricator 和我们的代码管理系统；<\/li>\n\t<li>最后对 Facebook 最前沿的开发模式做一个展望。<\/li>\n<\/ul>\n","guestName":"黄力菲，谭映辉","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58f759d8a9af2.png","guestIntroduce":"0","beginTime":"1499502900","endTime":"1499505600","address":"马德里2号厅","link":"","extension":{"spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d24874d1e56.png","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1489413375","updatedAt":"1489413375","status":"-1","days":"08","spec_topic_pic":"http:\/\/static.geekbang.org\/ck\/58d24874d1e56.png","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"Facebook","guestPosition":"Engineering Manager，开发工具资深专家","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/974\/5960980c14ef4.pdf\/18"}],"producter_name":"葛俊","topic_id":"85","order":"44","topic_description":"<p>Donald Knuth 为了出版他的 Art of Computer Science 设计并实现了 Tex；<\/p>\n\n<p>Linux Core 团队为了更好支持其项目代码管理研发了 Git；<\/p>\n\n<p>Facebook、Google、Amazon 等公司都有业界闻名的工具和工具团队。<\/p>\n\n<p>强大的工具对个人、团队和公司的开发效率都有着巨大的提升，对开发者的技术成长有非常积极的促进作用，更重要的，它让开发工作变得更愉快！<\/p>\n\n<p>在这专题里我们给大家介绍一些世界知名公司使用工具的经验，看看他们如何使用工具来提高效率。更为重要的是，我们将探讨互联网时代开发工具背后的出发点、设计思路和具体实现办法，为面临愈发激烈竞争的互联网公司在提高效率方面提供参考。<\/p>\n","introduce":"本专题里为大家介绍世界知名公司使用工具的经验，我们将探讨互联网时代开发工具背后的出发点、设计思路和具体实现办法，为面临愈发激烈竞争的互联网公司在提高效率方面提供参考。","guestavatar":"http:\/\/static.geekbang.org\/ck\/590757ffafe85.png","company":"华为","position":"技术专家"},{"topic":"下一代视频直播和点播技术","content":[{"id":"956","conid":"18","topic":"下一代视频直播和点播技术","title":"快手在大数据驱动下的直播体验优化","introduce":"<p>移动端视频直播业务经过2016年的井喷期，已经进入下半场，大家的关注点已经从如何构建完善的直播平台的粗放增长阶段，转入精细化运营阶段。如何在巨大的流量、复杂的应用场景、复杂的网络条件下，持续优化用户体验，是我们亟待回答的问题。<\/p>\n\n<p>构建大数据驱动的直播优化体系是快手为应对这一难题所提出的解决方案。为此，我们设计了各个层次，各个维度的数据监控指标，并利用成熟的大数据平台进行数据分析和可视化。<\/p>\n\n<p>一切优化均围绕着数据进行：找到用户痛点，指导优化方向，评估优化效果，走出了一条行之有效的数据驱动之路。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>快手直播的业务背景，移动直播体验优化的方法论；<\/li>\n\t<li>面向用户体验质量 (QoE) 和服务质量 (QoS) 的数据监测指标体系设计；<\/li>\n\t<li>系统构建历程和真实优化案例。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解如何针对大规模移动直播平台进行质量监测；<\/li>\n\t<li>深入浅出的理解直播监测指标的设计，以及如何做有针对性的优化；<\/li>\n\t<li>理解技术架构方案的选择：大数据平台的搭建，可视化、监控方案等。<\/li>\n<\/ul>\n","guestName":"罗喆","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/595b3925d6fae.png","guestIntroduce":"0","beginTime":"1499491800","endTime":"1499494500","address":"巴塞罗那厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1493703418","updatedAt":"1493703418","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"快手","guestPosition":"软件工程师","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/963\/59607f0ebc0db.pdf\/18"},{"id":"963","conid":"18","topic":"下一代视频直播和点播技术","title":"Hulu基于DASH构建的高清直播系统架构及实践","introduce":"<p>Hulu 作为美国三大在线专业视频服务提供商之一，不满足于提供单纯的点播视频服务，于2017年五月推出了自己的电视直播服务。<\/p>\n\n<p>Hulu 依靠自己运营高清点播服务的经验和基础架构，重新搭建了整个直播系统，并对现有的点播系统进行了全面的升级。该架构适应了 Hulu 互联网电视产品的需求，可以为从移动端到客厅大屏各种设备提供高清、快速响应的直播／点播视频体验。<\/p>\n\n<p>相比国内直播应用，Hulu 的直播系统有以下的一些特殊需求：<\/p>\n\n<p>&nbsp; 1. 高稳定，高可扩展：需支持24x7无停机直播，随时应对数千路直播上直播节目带来用户的&ldquo;惊群&rdquo;问题；<\/p>\n\n<p>&nbsp; 2. 高质量：视频质量最高可达1080p，60fps。需要达到与有线电视近似的延迟（5至10秒）。并可随时监控观看质量变化；<\/p>\n\n<p>&nbsp; 3. 复杂的商业逻辑：需要考虑用户的各种状态（地理位置，并发观看数等等）以及视频节目信息，随时控制用户播放。<\/p>\n\n<p>与传统 IPTV 相比，Hulu 的直播系统又是一个基于互联网的直播系统，需要适应互联网本身的一些特性，如网络波动，终端类型繁杂等。<\/p>\n\n<p>本次分享将介绍 Hulu 直播系统的一些创新之处，并介绍我们在提高直播系统稳定性及可扩展性，降低直播延迟，提高直播视频质量，方面的一些实践经验。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>Hulu 的直播产品简介（主要介绍 Hulu 新直播产品的核心功能及技术需求）；<\/li>\n\t<li>基于 DASH 的直播核心系统的架构（主要介绍从直播流接入，到转码，封装加密，CDN 传输直到终端播放的直播核心系统架构）；<\/li>\n\t<li>直播体验优化；\n\t<p style=\"margin-left:2px\">a. 直播延迟控制<\/p>\n\n\t<p style=\"margin-left:2px\">b. 视频质量跟踪与改善<\/p>\n\n\t<p style=\"margin-left:2px\">c. 系统稳定性及可扩展性<\/p>\n\t<\/li>\n\t<li>下一步的展望。<\/li>\n<\/ul>\n\n<h2>&nbsp;听众受益<\/h2>\n\n<ul>\n\t<li>了解美国视频业界最新的产品和技术动向；<\/li>\n\t<li>了解最新视频格式 DASH 及相关技术发展；<\/li>\n\t<li>了解搭建高清直播／点播系统的实战经验。<\/li>\n<\/ul>\n","guestName":"李彬","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/590c18af912ef.png","guestIntroduce":"0","beginTime":"1499495400","endTime":"1499498100","address":"巴塞罗那厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1493952734","updatedAt":"1493952734","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"Hulu","guestPosition":"研发中心首席软件开发主管","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/967\/59608bdd34da4.pdf\/18"},{"id":"1019","conid":"18","topic":"下一代视频直播和点播技术","title":"YY直播基于软硬件的弱网深度优化","introduce":"<p>随着户外直播等形式的兴起，采用手机在多变的网络环境下进行直播已成为常态，与之而来的是高丢包、低带宽等因素而导致的质量不稳定问题。<\/p>\n\n<p>本次分享介绍了 YY 直播针对质量较差网络（简称弱网）的环境，基于数据分析，在客户端和云端所采取的一系列技术手段。<\/p>\n\n<p>同时，就如何改善上下行网络环境，也给出自己的一些解决方案。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>移动直播面临的挑战；<\/li>\n\t<li>弱网的软件解决方案；<\/li>\n\t<li>弱网的硬件解决方案；<\/li>\n\t<li>案例综述。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解影响直播质量的各种网络相关问题；<\/li>\n\t<li>了解弱网的解决方案及其背后的逻辑和原理；<\/li>\n\t<li>了解 YY 直播平台相关技术。<\/li>\n<\/ul>\n","guestName":"林正显","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/592e1fdf63da0.png","guestIntroduce":"0","beginTime":"1499499300","endTime":"1499502000","address":"巴塞罗那厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1496194075","updatedAt":"1496194075","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"欢聚时代（YY）","guestPosition":"研发总监","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/973\/5960959eca1af.pdf\/18"},{"id":"1041","conid":"18","topic":"下一代视频直播和点播技术","title":"QQ空间亿级社交短视频播放实战","introduce":"<p>近两年短视频发展迅猛，QQ空间作为国内最大的社交平台，如何将短视频从百万级播放做到10亿级播放，架构上如何支撑，体验／性能／成本问题如何解决，这里和大家分享下我们的一些实践。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>从0到1构建<\/li>\n\t<li>体验／性能／成本优化<\/li>\n\t<li>进阶思考<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解QQ空间的短视频技术架构<\/li>\n\t<li>参考QQ空间的短视频实践经验<\/li>\n<\/ul>\n","guestName":"王磊","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5934e196a9ed5.png","guestIntroduce":"0","beginTime":"1499502900","endTime":"1499505600","address":"巴塞罗那厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1496637973","updatedAt":"1496637973","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"腾讯","guestPosition":"QQ空间iOS技术总监","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/977\/5960a33d04c89.pdf\/18"}],"producter_name":"洪小军","topic_id":"86","order":"43","topic_description":"<p>视频直播在经历2015和2016年的发展和井喷期后，开始进入直播下半场，技术也进入更加体系化和精细化的阶段。大数据驱动的优化、弱网层面体系化优化等都是这个阶段很重要的部分及其往下一阶段发展的关键突破点。随着直播形态逐步被更多人接受，传统的一些业务形态也在与直播做结合，比如高清互联网电视产品结合直播是怎么样的体验，这里会面临哪些新的挑战，及其对于现有的移动直播会有哪些可借鉴的点，这些也是大家在不断思考和探索的方面。<\/p>\n\n<p>短视频在近几年大放异彩，各大巨头纷纷在加大布局，用户在短视频应用方面的时间投入也不断增多。视频点播技术作为一项更加成熟的技术，在新的发展阶段，亿级以上播放、成本和体验期望得到更好的保障等，这其中如何去更好支撑。<\/p>\n\n<p>结合业内典型应用场景的一线实践案例，期望通过本专题来跟大家一起探讨下一代视频直播和点播技术的最佳实践和发展趋势。<\/p>\n","introduce":"视频直播在经历2015和2016年的发展和井喷期后，开始进入直播下半场，技术也进入更加体系化和精细化的阶段。大数据驱动的优化、弱网层面体系化优化等都是这个阶段很重要的部分及其往下一阶段发展的关键突破点。随着直播形态逐步被更多人接受，传统的一些业务形态也在与直播做结合，比如高清互联网电视产品结合直播是怎么样的体验，这里会面临哪些新的挑战，及其对于现有的移动直播会有哪些可借鉴的点，这些也是大家在不断思考和探索的方面。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58b788ff8809d.png","company":"美图","position":"技术总监"},{"topic":"区块链以及金融新技术","content":[{"id":"1000","conid":"18","topic":"区块链以及金融新技术","title":"点融网金融科技的大数据解决方案","introduce":"<p>大数据应用是承载和推动金融科技行业（Fintech）这些年快速发展的重要因素，但如 Fintech 自身的创新性和探索性一样，大数据在该领域的应用也需要通过大量的实践结合业务规模和发展探寻出一条适合的道路。<\/p>\n\n<p>本次分享将结合点融网近些年在大数据架构上的探索，介绍如何根据业务需求设计改造升级了当前的一整套架构。从介绍大数据可以在金融科技的哪些场景下发挥极大的作用开始，接下来会深入介绍针对这些场景，如何设计和实施合适高效的数据产品。<\/p>\n\n<p>同时这次分享会包含大量的数据产品经验分享如数据的搜集、清洗、存储、处理、机器学习建模等知识，以及这些大数据产品如何回馈于现有的业务。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>Fintech 的场景介绍&nbsp;<\/li>\n\t<li>借款人数据系统设计&nbsp;<\/li>\n\t<li>贷款人数据系统设计<\/li>\n\t<li>智能商务逻辑解决方案&nbsp;<\/li>\n\t<li>整体大数据架构<\/li>\n<\/ul>\n","guestName":"单忆南","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/591bfafb823f3.png","guestIntroduce":"0","beginTime":"1499491800","endTime":"1499494500","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1495009591","updatedAt":"1495009591","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"点融网","guestPosition":"大数据高级总监","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/962\/5960775bdc1b8.pdf\/18"},{"id":"967","conid":"18","topic":"区块链以及金融新技术","title":"区块链技术及其在金融领域的落地实践","introduce":"<p>区块链作为一种可信的价值载体，将对金融行业产生极大的影响，如何将区块链的技术特征和金融业务结合在一起，成为一个大家都很关注的话题。本次分享将围绕区块链相关技术在传统金融银行领域的技术实践经验，介绍区块链技术如何在业务中落地应用。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>区块链技术简介及其在金融领域的应用场景；<\/li>\n\t<li>Hyperledger Fabric ;<\/li>\n\t<li>区块链技术在银行的落地实践；<\/li>\n<\/ul>\n","guestName":"韩昭芳","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/591289d8cba4a.png","guestIntroduce":"0","beginTime":"1499495400","endTime":"1499498100","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1494320030","updatedAt":"1494320030","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"北京阿尔山金融","guestPosition":"技术总监","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/965\/5960867fd3e9e.pdf\/18"},{"id":"968","conid":"18","topic":"区块链以及金融新技术","title":"58到家支付系统架构与实践","introduce":"<p>本议题介绍58到家目前支付系统的业务架构、技术架构、业务数据监控等，目前到家一共有速运业务线、保洁业务线、保姆业务线、平台业务线、月嫂业务线等五大业务线，各业务线的支付流程都有一些共同点和不同点。<\/p>\n\n<p>我们统一的支付系统需要灵活地对接各业务线并保证支付流程和数据的统一，到家支付体系目前由收银台、交易、支付、对账、清结算、网关、账户、基础服务、实时监控、业务报表等各系统组成。<\/p>\n\n<p>在几年的演进过程中我们遇到过哪些问题？怎么解决的问题？一个完善的支付体系需要注意哪些事情？系统经过这几年的演进和优化，我们针对高并发访问、数据一致性保证、灵活的业务支撑相比之前都强大很多，确保了互联网金融系统中资金数据的真实性、实效性、完整性、准确性和可验证性。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>58到家支付系统介绍；<\/li>\n\t<li>支付系统业务架构；<\/li>\n\t<li>支付系统逻辑架构；<\/li>\n\t<li>业务数据监控和系统监控；<\/li>\n\t<li>对账系统；<\/li>\n\t<li>遇到的问题和相关实践。<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>了解支付系统的整体架构；<\/li>\n\t<li>了解收银台、交易、支付、对账、清结算、网关、账户、基础服务、实时监控、业务报表等系统；<\/li>\n\t<li>了解支付系统是如何保证资金数据的真实性、实效性、完整性、准确性和可验证性；<\/li>\n\t<li>58到家实践遇到的一些问题和经验。<\/li>\n<\/ul>\n","guestName":"蔡敏","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5911812066e4a.png","guestIntroduce":"0","beginTime":"1499499300","endTime":"1499502000","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1494320208","updatedAt":"1494320208","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"58到家","guestPosition":"架构师&支付平台部技术负责人","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/968\/5960921c6e78c.pdf\/18"},{"id":"882","conid":"18","topic":"区块链以及金融新技术","title":"支撑千万亿级交易额的银行云计算架构演进","introduce":"<p>以可用性为第一目标的金融在线服务在过去的20年内完成了省级集中、大集中、容灾中心、两地三中心、双活数据中心到云的多年发展，在移动互联网和云计算为 IT 基础的今天，银行如何在服务千万亿交易额的业务同时，考虑到成本、快捷和提升用户体验，真正实现可用、易用和好用。<\/p>\n\n<p>本次演讲内容将由银行数据中心架构到云计算演进的设计建设过程，运行维护和问题的快速发现处理为主线，为大家讲述银行数据中心建设的过去，现在和未来3～5年的预期。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>银行数据中心发展历史；<\/li>\n\t<li>互联网+，Fintech 大潮下给银行数据中心建设带来的挑战；<\/li>\n\t<li>云+端架构在银行业的设计、实现和最佳实践。<\/li>\n<\/ul>\n","guestName":"龙成","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/594a46cd3e2ac.png","guestIntroduce":"0","beginTime":"1499502900","endTime":"1499505600","address":"马德里3号厅","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1491481151","updatedAt":"1491481151","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"听云","guestPosition":"金融行业系统架构师","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/975\/5960a2c9d7d0e.pdf\/18"}],"producter_name":"巨建华","topic_id":"78","order":"47","topic_description":"<p>过去几年以来，云计算和大数据为 FinTech 的快速发展提供了坚实的基础，使企业可以用极低的成本专注于业务创新。<\/p>\n\n<p>人工智能和区块链则为 FinTech 插上了腾飞的翅膀，使技术真正成为了改变行业竞争格局的利器，受到了前所未有的关注。<\/p>\n\n<p>在突飞猛进的技术发展浪潮中，如何驾驭这些前沿技术并推动业务变革是当前架构师们最大的挑战。<\/p>\n\n<p>本专题将邀请处于技术浪潮一线的架构师，现身分享 FinTech 在实际业务中的实践经验，为听众打开通往技术应用创新之门。<\/p>\n\n<p>&nbsp;<\/p>\n","introduce":"在突飞猛进的技术发展浪潮中，如何驾驭区块链等前沿技术并推动业务变革是架构师们最大的挑战，本专题将邀请处于技术浪潮一线的架构师，现身分享 FinTech 在实际业务中的实践经验。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58b789b9d7e9a.png","company":"小赢科技","position":" 技术副总裁"},{"topic":"电商之核心架构","content":[{"id":"1012","conid":"18","topic":"电商之核心架构","title":"苏宁高并发库存系统架构演进及实践","introduce":"<p>库存系统历来是电商平台的心脏，不但要面对高并发的查询场景，更是整条交易链路上资源争抢的核心所在，其性能高低、可用性好坏、扩展性强弱直接决定了电商平台的高性能、高可用、高可扩展性。又因其数据一致要求非常之高，所以设计实现过程中难点非常之多，挑战性很大。<\/p>\n\n<p>苏宁电商业务又有O2O双线发展、业务场景销售模式繁多、完善的自营模式等等特点，库存业务的复杂性较普通平台式电商要高的多。在此背景下，库存产品线进行了一系列优化升级，近三年来进行了大刀阔斧的架构变革，使库存系统性能、可用性、稳定性实现了质的飞跃，稳定支撑了业务量大幅攀升。<\/p>\n\n<p>此次分享，以苏宁库存交易系统的架构演进为主线，详细围绕苏宁近10年信息化各阶段中库存遇到的问题及架构改进思路，并介绍库存设计的架构最佳实践。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>库存业务介绍及挑战：\n\t<p style=\"margin-left:2px\">a. 库存系统定位<\/p>\n\n\t<p style=\"margin-left:2px\">b. 核心场景分析<\/p>\n\n\t<p style=\"margin-left:2px\">c. 库存业务架构<\/p>\n\n\t<p style=\"margin-left:2px\">d. 架构演进概览<\/p>\n\t<\/li>\n\t<li>电商初期库存架构<\/li>\n\t<li>前台、中台、后台分离架构<\/li>\n\t<li>自研库存架构\n\t<p style=\"margin-left:2px\">a. 总体架构<\/p>\n\n\t<p style=\"margin-left:2px\">b. 系统交互<\/p>\n\n\t<p style=\"margin-left:2px\">c. 抢购库存架构<\/p>\n\t<\/li>\n\t<li>多活库存架构\n\t<p style=\"margin-left:2px\">a. 单元化设计原则<\/p>\n\n\t<p style=\"margin-left:2px\">b. 库存多活设计<\/p>\n\t<\/li>\n<\/ul>\n","guestName":"司孝波","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5927ddf889a32.png","guestIntroduce":"0","beginTime":"1499491800","endTime":"1499494500","address":"大宴会厅1","link":"","extension":{"spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0"},"createdAt":"1495534650","updatedAt":"1495534650","status":"0","days":"08","spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0","guestCompany":"苏宁云商","guestPosition":"IT总部技术总监","dayorder":2,"ppt_link":""},{"id":"827","conid":"18","topic":"电商之核心架构","title":"业务规模驱动技术升级：京东物流系统高并发架构演进之路","introduce":"<p>物流是电商重要的一环，近几年随着电商爆发式的增长，订单量的成倍增长和客户对物流体验的极致要求对电商物流系统来说都是极大的冲击和考验。<\/p>\n\n<p>在业务规模的驱动之下，京东物流系统经历了怎样的架构演进之路？面对亿级单量，系统又做了哪些应对高并发的技术升级？<\/p>\n\n<p>京东物流系统经过多年的技术升级与大促考验，足以支撑亿级单量的顺畅生产，这次分享将聚焦物流系统高并发技术架构，旨在介绍京东物流系统在高并发架构方面的演进过程。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>物流系统关键词<\/li>\n\t<li>物流系统高并发场景<\/li>\n\t<li>静态数据的缓存之路<\/li>\n\t<li>垂直和水平拆分<\/li>\n\t<li>分离技术<\/li>\n\t<li>+互联网到DB+<\/li>\n<\/ul>\n\n<h2>听众受益<\/h2>\n\n<ul>\n\t<li>京东物流系统的基本架构；<\/li>\n\t<li>京东物流系统应对高并发的一些架构模型和优化经验；<\/li>\n\t<li>在业务规模的驱动之下，京东仓储和配送系统的技术架构演进过程。<\/li>\n<\/ul>\n","guestName":"者文明","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/58e735b055290.png","guestIntroduce":"0","beginTime":"1499495400","endTime":"1499498100","address":"大宴会厅1","link":"","extension":{"spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0"},"createdAt":"1490606082","updatedAt":"1490606082","status":"0","days":"08","spec_topic_pic":"","ppt_status":"1","video_status":"0","overseas":"0","guestCompany":"京东商城","guestPosition":"运营研发部总架构师","dayorder":2,"ppt_link":""},{"id":"1015","conid":"18","topic":"电商之核心架构","title":"Walmart eCommerce search system architecture: intro and evolution","introduce":"<p>As one of core parts in Walmart eCommerce platform, search system is facing challenges like fast growing number of products, increasing search traffic volume, real time update, and new business needs. To handle these challenges, we continuously improve and redesign the search architecture.&nbsp;<\/p>\n\n<p>This session will first introduce the architecture of Walmart eCommerce search system, followed by its challenges; then discuss several architecture design cases with focus on the search backend system; finally share experience and lessons we have learned in this evolution.&nbsp;<\/p>\n\n<h2>Agenda&nbsp;<\/h2>\n\n<ul>\n\t<li>Introduce Walmart eCommerce search system&nbsp;<\/li>\n\t<li>List challenges in the past several years&nbsp;<\/li>\n\t<li>Show the search architecture evolution&nbsp;<\/li>\n\t<li>\n\t<p style=\"margin-left:2px\">a. Distributed search cloud (vs. open source Solr Cloud)&nbsp;<\/p>\n\n\t<p style=\"margin-left:2px\">b. Re-rank (decouple from underlying search engine)&nbsp;<\/p>\n\n\t<p style=\"margin-left:2px\">c. Metadata store (support real time update)&nbsp;<\/p>\n\n\t<p style=\"margin-left:2px\">d. Microservices&nbsp;<\/p>\n\t<\/li>\n\t<li>Lessons learned&nbsp;\n\t<p style=\"margin-left:2px\">a. Pros &amp; Cons using open source&nbsp;<\/p>\n\n\t<p style=\"margin-left:2px\">b. Chaos by microservices Audiences&nbsp;<\/p>\n\t<\/li>\n<\/ul>\n\n<h2 style=\"margin-left:2px\">get to know<\/h2>\n\n<ul>\n\t<li>When to adopt\/customize\/replace open source software architecture<\/li>\n\t<li>Evolution experience with challenges of business growth<\/li>\n\t<li>The architecture of Walmart eCommerce search system&nbsp;<\/li>\n<\/ul>\n","guestName":"曹宁","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/59253cc381343.png","guestIntroduce":"0","beginTime":"1499499300","endTime":"1499502000","address":"大宴会厅1","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1"},"createdAt":"1495613412","updatedAt":"1495613412","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"1","guestCompany":"WalmartLabs","guestPosition":"Engineering Manager","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/972\/596095649ac63.pdf\/18"},{"id":"1004","conid":"18","topic":"电商之核心架构","title":"阿里异地多活与同城双活的架构演进","introduce":"<p>为什么我们要做异地多活？简单来说，一是因为双11很火（扩容的需要），二是因为杭州很热（容灾的需要），这两个需求在各个层面都影响着阿里巴巴各项业务的飞速发展。<\/p>\n\n<p>为了不影响业务的调整发展，阿里巴巴通过&ldquo;异地多活&rdquo;技术，将业务封闭在一个个逻辑单元内，然后在多地部署多个这样的逻辑单元，满足了业务对容量和容灾的需求。同时由于有个别业务对&ldquo;中心单元&rdquo;有强依赖，我们又通过中心的同城双活解决了容灾问题。<\/p>\n\n<h2>演讲提纲<\/h2>\n\n<ul>\n\t<li>阿里做多活的背景、难点及方案选型；<\/li>\n\t<li>阿里从本地两个小单元 POC 到千里之外的&ldquo;三地五单元&rdquo;的多活的架构演进历程；<\/li>\n\t<li>在异地多活的架构下，同城双活的难点挑战及发展历程；<\/li>\n\t<li>阿里的异地及同城容灾切换能力及容量的水平伸缩能力。<\/li>\n<\/ul>\n","guestName":"谢吉宝（唐三）","guestTitle":"0","guestAvatar":"http:\/\/static.geekbang.org\/ck\/5923da875d177.png","guestIntroduce":"0","beginTime":"1499502900","endTime":"1499505600","address":"大宴会厅1","link":"","extension":{"spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0"},"createdAt":"1495440606","updatedAt":"1495440606","status":"0","days":"08","spec_topic_pic":"","ppt_status":"0","video_status":"0","overseas":"0","guestCompany":"阿里巴巴","guestPosition":"高级技术专家","dayorder":2,"ppt_link":"http:\/\/ppt.geekbang.org\/slide\/download\/976\/5960a303d7a49.pdf\/18"}],"producter_name":"孙迁","topic_id":"89","order":"41","topic_description":"<p>电商的竞争最终是技术的较量，电商是各大热门技术的集大成者，但电商系统架构没有标准模式模板。<\/p>\n\n<p>电商公司都是在各自发展的过程中逐步摸索和踩坑适合自己业务高速发展的技术和业务架构。<\/p>\n\n<p>核心架构演进过程中会遇到哪些痛点？让我们听一听主流电商公司系统架构发展过程中的宝贵经验。<\/p>\n","introduce":"电商的竞争最终是技术的较量，电商是各大热门技术的集大成者，本专题将从电商架构出发，探讨近年发展趋势、新应用落地以及创新实践等各种核心设计。","guestavatar":"http:\/\/static.geekbang.org\/ck\/58b791a4452a5.png","company":"苏宁云商","position":"IT总部总监"}]}