Technological Innova.on at Alibaba Alan Qi Vice President of Ant Financial Service Group
Outline 11-11 & Alibaba Technologies at Alibaba Conclusions
What is Singles’day (11-11)?
Black Friday vs. 11.11 Black Friday ( 2014) 11.11 Shopping Festival ( 2015) 1 Country 1 Day 1 Company 1 Day 2.3 Billion 14.3 Billion Dollar Online GMV 9.1 Billion Dollar Online GMV Dollar Offline GMV >249M 249M Online Customer Online & Offline Customer 27.9% 68% On Mobile Client On Mobile Client
“11-11”sales revenue explodes GMV In 7 years, sales revenue 600 increases 154 times 571 500 400 2015/11/11: 91.2 Billion 350 Yuan (14.3 Billion USD) 300 200 191 100 5.9 19.4 53.3 0 2009 2010 2011 2012 2013 2014 Unit: 0.1 Billion Yuan
“11-11”number of orders explodes ( 10K) In 6 years, number of orders 30000 27800 increases 57 times 25000 18800 20000 2014/11/11 15000 Total orders: 278 Millions 10000 10000 3369 5000 1280 490.00 0 2009 2010 2011 2012 2013 2014 IOE architecture Cloud
“11-11”order peak value explodes ( per second ） 90000 In 6 years, order peak values 80000 increases 80000 200 times 70000 60000 50000 42000 2014/11/11 Order peak value: 80K per 40000 sec. 30000 20000 14000 10000 3200 1000 400 0 2009 2010 2011 2012 2013 2014 IOE architecture Cloud
“11-11”transcation peak value explodes (per second ） 45000 In 6 years, transaction peak 38500 40000 values increases 193 times 35000 30000 25000 2014/11/11 20000 Alipay: 38.5K per sec. 15000 15000 10000 3800 5000 1200 500 200 0 2009 2010 2011 2012 2013 2014 IOE architecture Cloud
“11-11”transcation peak value explodes (per second ） 45000 In 7 years, transaction peak 38500 40000 values increases 430 times 35000 30000 25000 2015/11/11 20000 Alipay: 85.9K per sec. 15000 15000 Visa: 56K per sec. in lab 10000 14K per sec in real 3800 5000 1200 Master: 40K per sec. in lab 500 200 0 Alipay: Strongest 2009 2010 2011 2012 2013 2014 processing power among all payment systems IOE architecture Cloud
Capacity of processing financial big data explodes Data processing capacity （ PB ） In 6 years, data processing capacity increases 12000 100K 10240 10000 times 8000 6000 2014/11/11： 4000 Total processed data： 100PB 2000 1024 100 0.10 10 30 0 2009 2010 2011 2012 2013 2014 Oracle RAC Hadoop ODPS data Single BU Alibaba Data from data Group Ecosystems
“11-11”needs information security In 11/11/2014，Alibaba successfully defended 200 DDoS attacks; the biggest attack volume is bigger than 100Gbps.
The company behind 11-11?
Alibaba History Alibaba Group is reorganized into 25 Taobao is re- organized into business units; announces RMB1M profit Alibaba founded in Jack a plan to set up Small and Taobao Market- Alibaba announced daily; Alipay is Ma’s apartment in place, Tmall.com Micro Financial Services “Hangzhou launched Alibaba Cloud Hangzhou and eTao Group; leads the formation +Beijing” dual- Company breaks even Alibaba.com lists in HK Computing is founded of Cainiao center strategy 1999 2002 2004 2007 2009 2011 2013 2015 2001 2003 2005 2008 2010 2012 2014 September 19 th : Alibaba Group establishes AliExpress is launched; Tmall.com is introduced Alibaba.com the Alibaba Foundation Alibaba Group’s IPO RMB1M tax paid Juhuasuan and eTao are surpasses 1M with a sizeable fund (Nasdaq: BABA) RMB1M revenue daily; daily; Acquired introduced registered users dedicated to social Taobao is founded China Yahoo! causes.
Alibaba Strategy e-Commerce Health Entertain- ment Data Finance Logistics
Alibaba Ecosystem Logistics Infrastructure O2O Social Online Digital Network Other Entertainment Marketing Participants Ant Financial Logistics UC Browser Partners buyer sellers DATA Alibaba e-commerce infrastructure
Technologies at Alibaba
Key Technologies Deep user profiling & interest mining 9 large data centers. to better connect advertisers to their Active-active data centers across target audience multiple cities technology Large-scale real-time “search by Break 4 world records on Sort image” technology that powers r Benchmark 2015. camera-based product search 3 time faster than Apache Spark A secure e-commerce ecosystem for online transactions, data Dynamic restoration & exchanges, cloud computing and deployment ability of mobile clients mobile interactions. Super-large scale learning from data with billions of attributes and Ten billion level RPC and message billions of samples processing capability
Technology challenges that Alibaba faced in 2009 存 High hardware cost IOE (IBM servers, Oracle databases and EMC storage): expensive for scalability. In a few years, the hardware cost could have bankrupted the whole company! Data islands 通 Data from different business units are located in different clusters. Extremely hard to share data with each other! No common data standard for usage 用 Due to isolated storage, data are moved between clusters with repeated storage and computations.
Milestones Public service ODPS platform 5K node in a single cluster Massive data processing platform went online at Alibaba financial service. 2010.04 2013.08 2015 2014.11 2009.09 2012.07 Big data platform & AliCloud Started CDO started to products have been fully integrate data tested by 11-11 19
Data-driven business at Alibaba Internet Finance Search & Customer ads for service ecom Dozens of business units Data business Hundreds of PB data Thousands of engineers Info Credit Security Intelligent logistics
Cloud Computing & Infrastructure
Aliyun 10/15/2015 Achieved ITU Telecom World Entrepreneurship Award for excellence in providing and promoting innovative ICT 08/15/2013 solutions with social impact 5000-node Asparas cluster went into production 08/27/2010 Apsaras became the 07/2014 platform of four 5K applications: Search, Launched ODPS, Mail, Image Storage, which is capable of 07/28/2011 AliFinance processing 100 petabytes of data in Aliyun.com went six hours. online, releasing 1 st cloud service: ECS 09/10/2009 Aliyun.com Inc established
World Records • Break 4 world records on Sort Benchmark 2015 • Sort 100 TB of data in 377 seconds, 3 times faster than the record set by Apache Spark in 2014
ODPS Capable of processing 100 petabytes of Distributed data in 6 hours. Online Data Analysis & Application System Warehouse Mining Safety Security External CDP Realtime CDP Realtime Stream SQL SQL External data batch batch data Oracle data data Easy to Use Oracle Mysql channel channel Mysql Text Text OTS Graph MR PAI … Tunnel Tunnel … SDK Authorization and Access Base Dev . Data Process Console A/B Test Platform Center
Public Service
Public Service CNTV: online resource integra.on for CCTV (biggest TV sta.on in China) and online service. Cost saving: >50% CNTV肩负对央视众多频道内容资源进行整合和转换通过互联网提 供在线服务
Reliable service via data sharing Active-active data centers across multiple cities We can distribute network traffic between our Shanghai, Hangzhou and Shenzhen data center without any interruption.
World-wide Data Centers 9 large data centers in China Mainland, Hong Kong, Singapore and America.
Green Data Center: Zhang Bei • Zhang Bei: Strong wind, averaged temperature < 37 degree F. • Deploy solar energy, wind power and cold water cooling solutions: 100% green • Estimated at least 45% energy saving for cooling.
Green Data Center: Thousand Island Lake 千岛湖 Cold lake water for cooling Energy Cut 10K PDU ton coal index: saving usage 1.3 90% time: No need for electricity
Information Security • Intercept 3 million deceptive transaction requests every day (user identification, etc.). Totally successful defense of DDoS attack ： 546,859GB • Totally successful defense of attack since 2014: 9,383,772,384 times • Dec. 20 th , 2014, Defeated a DDoS attack ， which lasted 14 hours and peak traffic reached • 453.8 Gb per second
Big Data & Machine Learning
11-11 customer service On 11/11/2015 • Total service volume: > 10 Millions • > 95%: supported by intelligent robots • < 5% : supported by human • >98% phone call: went through
AI at Alibaba Identification Credit & Risk Control Loans Intelligent Advertising… Customer Service Natural language Optimization Time series analysis Deep Bayes Learning Recommen dation Data Computing
Intelligent customer service § Speech recogni.on § Deep Learning for problem iden.fica.on & QA § Natural language processing § Knowledge base construc.on
Large-scale Machine Learning Tmall Real-time Taobao Mobile Search Ads Recommendation Search • Increasing CTR > 5% • Increasing CTR > 10% • Conduct online learning on • Speeding up model training • Boosting ad revenue Double - 11 Shopping more than 3 times significantly Festival in 2015 Deep Learning Online Learning Hashing etc. Parameter Server
Recommend
More recommend
