Data storage, data analysis, data security ... Now there are more and more topics around "data", which are getting closer and closer to people's lives.
From the unfamiliar to the familiar, the data is not only "out of the circle", but has even stood in the C position. Last year, the "Opinions on Building a More Perfect System and Mechanism for Factor Market Allocation" issued by the Central Committee clearly stated that data has become the fifth largest factor of production after land, labor, capital and technology.
After entering the information age, database, operating system and middleware, as the three most basic softwares of computers, support the normal operation of enterprises.
When data becomes a factor of production, it will inevitably usher in explosive growth, and the data storage and processing needs of enterprises will be further released. More importantly, the epidemic has accelerated the pace of digital transformation and accelerated the speed of enterprises going to the cloud.
From informationization to digitalization, changes in the times will always bring changes in the business world. How to use the database under the cloud native architecture has become a pain point for enterprises and an opportunity for cloud vendors. Werner vogels, CTO of Amazon AWS, has repeatedly stressed: "Database is the ultimate battle of cloud computing."
In the era of digital intelligence, what does cloud nativity mean? What are the core advantages of cloud native database over traditional database? Is it cloud native to move the database to the cloud? Based on these problems, Lei Feng.com started a dialogue with Li Feifei, vice president of Alibaba Group and head of Alibaba Cloud Database Products Division.
Domestic cloud native database, get rid of "the pain of the skin"
Nowadays, the business world of database is divided into two factions because of the emergence and development of cloud.
One is the traditional commercial database represented by Oracle, and the other is the cloud native database represented by AWS abroad and Alibaba Cloud at home, all of which are the products of "IOE Revolution".
In fact, there are three kinds of hot databases in the early days: hierarchical database, network database and relational database.
In the book "Top of the Tide", the author Wu Jun wrote such a view: "The rise of Oracle Bone Inscriptions largely depends on its first sight of the market prospect of relational databases, and it is superior to IBM in business model."
Therefore, before the cloud native database entered WTO, the world of database was always dominated by Oracle, and most domestic Internet companies had to adopt the mode of Oracle+IBM minicomputer +EMC to maintain normal operation.
The high cost makes the internet giants who have huge demand for databases "unbearable".
In 2009, the number of Oracle RAC cluster nodes in Alibaba reached a record 20. However, because Oracle does not have the function of flexible expansion, it can only buy minicomputers and databases according to peak traffic, which leads Ali to pay most of the profits brought by business increase to Oracle.
In the second year, Ali began to take the road of "IOE", built AliSQL based on open source MySQL, and successfully passed the test of Taobao Double 1 1. The domestic cloud native database officially got rid of the "skin pain" and was gradually recognized by the market.
On the other hand, foreign AWS announced Amazon Aurora, a self-developed database based on cloud computing, on 20 15. Aurora is a relational database, which can copy six copies of data across three available areas. Its biggest feature is high performance and high availability.
With the entry of cloud computing giants, cloud native database has become the mainstream at home and abroad step by step. According to Gartner's prediction, by 20021,the proportion of cloud databases in the whole database market will reach 50% for the first time, and by 2023, 75% of databases will run on the cloud platform.
As for the cloud native database, with the gradual out of the circle, it also makes people pay attention to the focus from "what is it?" Change it to "What problems can be solved?"
However, cloud native database has the problem of data island. If you can't access multiple data systems, enterprises will be "under great pressure" in data processing and data management, and even have data security risks.
Traditional data warehouses generally build off-line data warehouses based on T+ 1 data integration to support various analysis and services of enterprises. Traditional solutions will not only affect the stability of online business, but also difficult to support the real-time needs of enterprises.
Therefore, in Li Feifei's view, the cloud native database has reached the 2.0 stage. The problem to be solved at this stage is the pain point mentioned above.
On September 26th, at the Yunfeng Conference on Database Innovation in Alibaba Cloud, Alibaba Cloud released the first one-stop agile data warehouse solution. This scheme combines one-stop data management platform DMS and cloud native data warehouse AnalyticDB (ADB for short) to realize the technical architecture of warehouse integration, providing online data real-time warehousing, T+ 1 periodic snapshots, and on-demand warehouse opening. The data delay is as low as seconds, which keeps the business online and allows enterprises to release more value from online data.
Compared with traditional solutions, Alibaba Cloud one-stop agile data warehouse solution has four core advantages:
1, which has little influence on the business side, will not affect the normal operation of the business side because of data aggregation and real-time processing, and the CPU and memory occupation is less than 5%;
2. The transaction sequence and data accuracy are guaranteed, the processing link is short, and online data real-time processing and unloading are supported, with higher efficiency. The data transmission efficiency is 100m/s, and the data delay is within 10 second;
3. Support complex real-time data processing and calculation logic;
4. Low code operation can greatly reduce the construction difficulty of real-time warehouse, improve the construction efficiency, and support various real-time scenarios in the process of enterprise digital transformation.
In addition to real-time statistical analysis scenarios, enterprises also need to build periodic full snapshots to meet the needs of periodic data analysis.
The traditional general integration scheme of several warehouses will have a stable impact on the production business, and the timeliness of the general integration is poor, which can not meet the business needs of customers for data backtracking at any time.
Aiming at the periodic integration scenario of T+ 1, the one-stop agile data warehouse solution supports full-aperture data snapshot of T+ 1 based on zipper table, and users can generate full-aperture or incremental snapshots of each period according to their needs through simple steps.
In addition, the business can also carry out data backtracking at any time point according to needs, and quickly solve the problem of abnormal data.
When talking about the future development trend of the database, Li Feifei mentioned the following five points:
1, cloud native+distributed must be the standard of database, and distributed is already necessary. A distributed database consists of several interconnected databases, while a user-oriented database is displayed as a single database. Cloud native distributed database has the characteristics of ease of use, high scalability, rapid iteration and cost saving. From resource pooling to flexible expansion, to intelligent operation and maintenance, to offline integration, to solve the core demands of enterprise users.
2.AI for DB (database) and DB for AI will be the mainstream trend. More importantly, AI is used to make database operation and maintenance intelligent, especially under the premise of cloud native and distributed, because the database is not only flexible, available and extensible, but also the complexity of application and operation and maintenance is greatly reduced after deployment. In database, in the face of more and more unstructured data, analytical ability is very important.
3. In today's environment, the security and reliability of data become more and more important. How to ensure that the whole database system can provide encryption ability, multi-party secure computing ability and privacy protection ability in the process of processing data links is also a very important trend.
4. The ability of multimodal data processing will become more and more important. For example, the multi-modal processing ability of the new database will play an increasingly important role in the application scenarios such as labeling and smart battery prediction of new energy automobile enterprises.
5. One data, multiple data processing engines: realize warehouse integration, warehouse linkage, warehouse opening and seamless data flow.
The above judgment also reflects the trend of Alibaba Cloud database from the side, which is beyond doubt. But beyond that, the industry is most concerned about open source.
Over the past six months, many domestic manufacturers have put forward open source strategies, and the reasons behind them are obvious. In order to create an ecology. At this year's Alibaba Cloud Summit, Zhang Jianfeng, President of Alibaba Cloud Intelligent and President of Dharma Institute, summarized the key words of Alibaba Cloud's 20021year development as: good service, deep foundation, thick China and strong ecology.
Doing a good job in service and ecology has become the goal of today's manufacturers, and open source is the best choice.
When Ray Feng.com asked, "Will Alibaba Cloud Database open all its capabilities in the future?" When asked this question, Li Feifei gave the answer: "No."
The reason for this answer is that he has some judgments and opinions on open source.
Li Feifei said that these parts are commercial versions of the Alibaba Cloud database.
In fact, most database vendors in the industry will not open source their core capabilities, such as the core control components of TiDB and TiFlash.
Unlike database vendors like MongoDB, Cassandra, CouchDB, they started their business by open source. Open source is only the strategy of Alibaba Cloud database, not the lifeblood of Alibaba Cloud database.
A few years ago, some people in the industry said that when facing open source, domestic databases first need to solve problems such as trust and open source intellectual property rights. "Open source will make suppliers consider copyright and patents more seriously. In fact, after choosing open source, they put forward higher requirements for database vendors. "
Li Feifei believes that open source is only a choice, and the success of database open source does not mean that commercialization can be successful, nor does it mean that manufacturers are not advanced.
More precisely, open source is just an effective means.
Finally, Alibaba Cloud Database hopes that customers can quickly use the technology of Alibaba Cloud database products through the open source version and participate in the iterative process of technical products. In some high-level abilities, they can learn from the team's professional ability and Alibaba Cloud's service ability and become good business partners. These are some basic ideas about open source in Li Feifei and Alibaba Cloud databases. Leifeng net Leifeng net Leifeng net