Alibaba Cloud detailed the core technologies used during the record-breaking, 24-hour global shopping festival. The data intelligence backbone of Alibaba Group applied multiple mature innovations to power the cloud infrastructure and other key AI-enabled features, enabling millions of businesses, to reach hundred of millions of consumers worldwide.
Powerful cloud infrastructure supports extreme scalability and robustness
Launched during the Apsara Conference in September, the third generation of X-Dragon architecture fully supported the shopping event. The Elastic Bare Metal Servers improved cost efficiency by 20%. X-Dragon servers seamlessly integrate computing platforms, including the Elastic Compute Service (ECS) bare metal server and virtual machines, in a single architecture.
Tapping this strong foundation, Alibaba Cloud handled extreme scalability, processing USD1 billion of GMV (Gross Merchandise Volume) in the first 68 seconds of the event; volume hit USD10 billion of GMV in under 30 minutes. Overall, Alibaba Cloud’s infrastructure handled a record-shattering USD38.4 billion of GMV in 24 hours with zero downtime.
Combining advanced data processing and analytic capabilities with green technologies Apsara Operating System processed around 970PB of data to support the peak performance of 544,000 orders per second, setting a record during 11.11.
POLARDB, a database developed by Alibaba Cloud, supported sales activity with a peak of 87 million requests processed per second. The new generation cloud-native POLARDB database combines performance with compatibility and availability of traditional enterprise databases in a cost-effective manner.
Through self-developed green technologies, including liquid immersion and deep-water cooling, power consumption for every 10,000 e-Commerce transactions can be reduced to 2kWh. Overall data center energy consumption dropped 70% from last year, saving over 200,000 kWh on 11.11.
AI-enabled features to empower sellers from international brands to SMBs
Thousands of Tmall merchants were upgraded to Tmall Flagship Store 2.0 storefronts before 11.11. Version 2.0 offers a unique user interface with a 3D Taobao avatar for customers to try on clothes and jewelry, further engaging customers via rich, interactive content and an omnichannel experience. The conversion rate per purchase was increased by over 20% from last year, serving 1,167 key brands’ flagship stores this year.
Alime Shop Assistant, a customer-service chatbot supported by Alibaba DAMO Academy’s machine learning technologies, handled 97% of online customer inquiries on Alibaba’s ecommerce platforms Taobao and Tmall during 11.11. The virtual assistant also empowered hundreds of thousands of merchants on Alibaba’s platforms with intelligent customer services.
For example, Alime answered questions during live broadcasts, extracting answers from pictures and responding conversationally. The Alime chatbot service offers 11 languages to support communications between global buyers and sellers.
Alibaba DAMO Academy provides enhanced machine translation technology for cross-border ecommerce platform, AliExpress. The technology can help small and medium-sized merchants globally to automatically translate their store and product pages into 21 languages, including Russian, Spanish, Turkish, French and Arabic, for buyers across Europe, Asia, America and the Middle East.
During the shopping festival, the machine translation service was used 1.66 billion times, with over 200 billion words translated in different languages.
Better consumer experience with live streaming and voice shopping
During this year’s global shopping festival, transactions during Taobao Live generated nearly USD 2.8 billion (RMB20 billion) in sales. More than half of the live streaming Tmall merchants selling products such as cosmetics, clothing, food, cars and electronics experienced robust growth, with home décor products seeing 400% sales growth.
This year, Taobao developed a new, real-time communications framework for audio and video on Alibaba Cloud. Network latency was reduced to 1.5 seconds from five to seven seconds, improving real-time interaction experience between broadcasters and users. Video AI identified products mentioned by broadcasters during live streaming, and audiences could view product links simultaneously displayed onscreen.
During 11.11, over one million orders were placed and processed through voice command via Tmall Genie, Alibaba A.I. Labs’ smart speaker. Purchased items during the day included 810,000 eggs, 1.4 million tons of rice and 76 tons of liquid detergent, a sign that voice shopping has become an increasingly popular trend among Chinese consumers of all ages.