10 Years of Cloud Intelligence: Ground-Breaking Big Data Performance to Feed Advances in AI
The Evolution of Big Data and AI
Alibaba’s quest for big data computation started more than ten years ago, with our data processing platform, MaxCompute, that celebrated its tenth birthday in May 2019.
When Alibaba Cloud launched, it embraced the challenge of large-scale development, and MaxCompute was one of the first independently developed big data computing platforms in China.
Since then, the powerful platform has supported numerous industries, including e-commerce, agriculture, healthcare, finance and education, processing large amounts of data to serve society and consumers. The platform is continuously increasing in strength and capacity. For example, in the public cloud market, it is growing by 250 percent year-on-year in both user and computing capacity. MaxCompute comes into its own during our Double 11 Global Shopping Festival, when it processed more than 600PB of data every day and supported 120,000 transactions per second during peak online traffic hours.
Big Data Feeds AI Advances
MaxCompute is a key part of Alibaba Cloud’s offering, supporting 99 percent of big data storage and 95 percent of computing workloads in the group. This impressive capacity powers Artificial Intelligence (AI) and Machine Learning products such as City Brain — a service that utilizes real-time data to power breakthroughs in cities such as Hangzhou, Kuala Lumpur and Macao.
This shift towards Cloud Database Warehouse (CDW) providers was flagged by leading market research organization, Forrester, who recognized MaxCompute as one of the top CDW solutions globally, and the highest-ranked product from Asia. Forrester’s report comments that Alibaba Cloud’s combination of MaxCompute + DataWorks + ApsaraDB provides a reliable data analysis platform based on a wide range of infrastructures, meeting its customers’ demands for real-time and EB-level data processing capabilities.
MaxCompute has provided a solution that is helping many industries in their quest for digital transformation. It builds a stable, secure, and scalable data mid-end that enables efficient interaction with application systems, including end-to-end machine learning products such as Machine Learning for AI.
Alibaba Cloud’s Artificial Intelligence is achieving widespread recognition. For example, image recognition has become a key test of AI’s processing power in recent years, creating Internet memes when struggling to distinguish between a blueberry muffin and a Chihuahua or a curled-up puppy and a bagel. However, AI is making rapid progress in this area. In December 2018, Alibaba Cloud’s technical team topped the DAWNBench deep learning inference rankings. The competition has been developed by Stanford and ranks competitors in both inference performance and cost. The AI has been integrated into the acceleration framework of the Alibaba Cloud GPU computing platform, using images to improve the experience of GPU users.
Alibaba’s machine learning technology also succeeded in another well-known test: Microsoft’s Machine Reading Comprehension dataset. Alibaba’s AI beat humans in a reading comprehension test. DAMO Academy developed the deep cascade learning model that was a key factor in beating the challenge. The Natural Language Processing technology has already been applied to Alibaba Cloud’s products, including powering our customer service chatbot, Alime. This bot handled approximately 98% of requests made on Alibaba’s platforms including Taobao and Tmall during the 2018 Double 11 Global Shopping Festival.
Watch out for AI being integrated into further products and services, such as NLP working together with speech AI and machine translation to help people communicate in different languages in real-time. The challenge for users is to think about the potential applications of this incredible technology and how we can benefit from it.
Learn more about Alibaba Cloud’s journey in the past 10 years by visiting https://www.alibabacloud.com/campaign/10-year-anniversary