Mainland E-commerce giant Alibaba's dual 11 shopping festival, is a cloud processing, logistics deployment and network traffic management of the biggest challenges, in order to increase the demand while maintaining the stability of the system, computing technology is behind the scenes. According to Techwireasia reports, 2009 Alibaba shopping platform processing 400 orders per second, 2016 to 17,000, 2017-year double 11 shopping day peak processing up to 325,000 orders per second, an annual increase of 48%. As the business becomes more and more complex, it brings pressure on the computing ability, order payment ability, logistics arrangement and other infrastructures. To this end, Alibaba has invested in 2 key technology areas, namely, cloud computing management and artificial intelligence (AI). Enhancing user experience is the goal of introducing new technologies, including smooth experience and customized experience. Alibaba software engineers said that a few years ago everyone saw the same recommendations and products, now the platform can be more personalized, users can even see the recommendation of friends. The 2 main challenges facing Alibaba are to keep growing demand and keep it costs unchanged in the increasingly complex logistics system. The report noted that the Alibaba system has adopted more automation capabilities, especially automatic end-to-end load testing, to avoid the system infrastructure because of excessive demand and collapse, the loss of orders in vain. The End-to-end tool for the load test system includes anonymous simulations of all system traffic data, and traffic data is the user record of Alibaba over the years. Alibaba engineers repeatedly simulate traffic on different load standards, and during the testing process, the machine intelligence and automatic scaling system will respond to these loads, adjust resource allocations, and ensure system stability. If the traffic is at a peak, the machine algorithm will transfer any remaining capacity to payment and order requirements, and some unnecessary bulk analysis can be quickly degraded to facilitate space, which is operated by Alibaba's dispatch Ai ' da Ling '. Because the system can operate autonomously, it can reduce the burden of engineers. As Alibaba's infrastructure is increasingly automated, it is reducing its reliance on manpower. The report points out that the online system itself, including automatic scaling, collaborative positioning technology and other key tools are autonomous operation, or only a very small number of human-computer interaction. Network operation also integrates automatic error detection and recovery tools, to avoid the problem of traffic, so that the original 10 minutes of work, can be completed within 10 seconds. In this 2-year load test, Alibaba anticipates that the future manual preparation system work can save 1000 of hours. Production system automation, less human intervention, more reliable, the engineer vacated time can focus on improving machine intelligence, to do more effective use.
