Reference News Network reported on March 19 a few days ago, the global authoritative research organization Forrester Research (Forrester) released the "2018 First Quarter Cloud Data Warehouse" report. The report conducts a comprehensive evaluation of the main functions, regional performance, market segments and typical customers of big data service providers, and finally AWS, Alibaba Cloud, Google, and Microsoft have entered the global front-line camp. Alibaba Cloud became the only Chinese technology company selected.
The Forrester report is considered to be a guide for CIOs of major companies around the world to choose service providers, and is extremely authoritative in the industry. Cloud-based big data services have become a development trend in recent years due to their security, elastic scaling, fast deployment, and low cost. Correspondingly, on-premises big data analytics solutions are becoming obsolete.
Each vendor requested in this Forrester assessment should have the following: 1) mature big data warehouse products; 2) stand-alone big data warehouse solutions; 3) big data use cases; 4) public offerings ; 5) Regional market leadership; 6) Advanced technology;
Alibaba Cloud MaxCompute is the selected Chinese product, and Forrester analyzed it in detail in the report.
Next, we will take the lead in publicly interpreting MaxCompute, the Alibaba Cloud big data computing service selected this time.
1) In terms of product maturity
In 2009, Alibaba encountered the ceiling of Greenplum. Greenplum is difficult to expand after 1000 machines and 1000TB, but even the scale of 100 machines is far from enough for a booming enterprise like Alibaba.
In September 2009, Alibaba Cloud started the research and development of the Feitian big data platform, with the vision to build a set of self-developed EB-level data warehouse MaxCompute.
2) Product function
Cloud-based big data services have become a development trend in recent years due to their security, elastic scaling, fast deployment, and low cost. Last year Forrester pointed out that although CDW enterprises have good cloud features, many cloud vendors also expose some deficiencies, such as service globalization, data security, integration, modeling and governance;
MaxCompute has always focused on globalization, performance, security, one-stop end-to-end development experience and further improvements in the ecosystem.
globalization
Currently, MaxCompute is located in 15 regions around the world, including Hong Kong, Singapore, Japan, Dubai, Europe, US West, US East, Australia, Indonesia, India, etc., connecting millions of servers into a supercomputer, providing online public services as a major global provider The Internet marketplace provides computing power.
performance
The EB-level performance processing of MaxCompute has reached the world's leading level. In October 2017, the world's first public cloud-based 100TB BigBench big data benchmark test was completed, and the performance exceeded 7830QPM.
Development efficiency
A new generation of big data language NewSQL, which combines the advantages of Declarative and Imperative, is proposed, which breaks through the technical limitations of the previous SQL language, unifies the programming language, supports offline, quasi-real-time, streaming, graph, machine learning and other computing modes and unstructured processing, greatly reducing the threshold for big data development.
Safety
The multi-tenant cloud security isolation technology is proposed, which breaks through the security limitations of traditional big data platforms and refines the security boundaries to the user, process, and kernel levels. It supports the authorization and auditing of millions of tenants and tens of billions of tasks per day, and realizes financial-level data security.
Data Modeling, Governance, Integration
Facing the demands of big data construction, management and application in all walks of life, MaxCompute provides one-stop big data capability tools for intelligent data construction and management from data access to data consumption, including DataWorks, MaxCompute Studio, etc., to help create standards A unified, integrated, capitalized, service-oriented, closed-loop self-optimizing intelligent data system to drive innovation.
3) Advanced technology
• In 2015, Feitian broke four world records for GraySort/MinuteSort in the Sort Benchmark ranking competition
– FuxiSort sort 100TB data in 377 seconds
– Shared-based test environment, dual Gigabit NICs, mechanical disks
• In 2016, Feitian broke two CloudSort world records in the Sort Benchmark ranking competition
– NADSort sorts 100TB of data for $144 ($1.44/TB)
– In 2017, MaxCompute adapted the TPC benchmark to expand the data scale to 100TB, and completed the world's first public cloud-based BigBench big data benchmark test. The performance exceeded 8200QPM, realizing China's computing and global capabilities.
• Spend $354.7/QPM
Qualification
• Passed 10 authoritative third-party industry/safety certifications
4) Regional market leadership
As the world's leading cloud computing vendor, Alibaba Cloud serves 2.3 million customers and covers more than 200 countries and regions around the world. Its public cloud market share is 47.6%, which is the total number of market followers. The big data computing service MaxCompute focuses on providing massive data storage and large-scale computing for customers in 15 regions around the world, including Hong Kong, Singapore, Japan, Dubai, Europe, US West, US East, Australia, Indonesia, India, etc. The computing power is exported to customers in various regions of the world.
5) Typical customer case
ofo yellow car
After using MaxCompute, ofo began to build data models and clustering, and learned the number of vehicles that need to be placed in various places based on historical orders and people flow information, where to pick up the cars, properly recycle data in cold areas, and increase delivery of data in hot areas. In July 2017, ofo upgraded from MaxCompute 1.0 to 2.0, increasing offline operation efficiency by more than 50%, easily handling 32 million daily orders, and increasing overall operation efficiency by 76%. At the same time, the operation and maintenance cost of the big data platform is greatly reduced, and only 0.5 people are required. Compared with the self-built physical cluster, the total cost of using Alibaba Cloud MaxCompute is greatly reduced, and the application development efficiency is greatly improved.
BGI
Genetic technology has gradually entered the life scene from the laboratory, and the volume of data has exploded, far beyond the scope that traditional computing power can support. Based on this background, BGI chose MaxCompute. In the Million Genomes Project, the traditional computing method takes 3-5 days to analyze the population structure. MaxCompute can complete the entire analysis within 1 hour, greatly accelerating the data throughput and delivery speed. When analyzing the genetic structure of millions of people's genetic data, the computational complexity makes traditional computing impossible. Using MaxCompute, BGI has made a technological breakthrough. It can complete the calculation of all genetic distances between one person and 100,000 people within a few hours. The cost has been greatly reduced to less than 1,000 US dollars. Such examples are still under development.
To sum up, Alibaba Cloud MaxCompute provides public cloud-based multi-tenant big data warehouses and hybrid cloud services. Globalization is progressing rapidly. Finance, Internet, retail, and e-commerce are its main business areas.
With its 9-year product maturity, technology leadership and one-stop big data development solutions, it has become the leader in the CBW cloud data warehouse market.