Domestic database chaos

author ： white eel
Source ： Cave of white eel ,CSDN Reprinted by authorization of the author to CSDN Information blog , To readers .

In fact, I started writing this article at the weekend , Writing this article this weekend , Many times I am thinking about a construction scheme of database localization and substitution , Read a lot of information . This year is my first year since I started working 31 A year , The first ten years of work , I've been writing code for ten years , From assembly 、COBOL To C Language , Hundreds of thousands of lines of code ; Over the next decade , I have been helping users make good use of the database , Also helping Oracle Extension RAC technology ;2015 Year begins , I continue to work on database optimization , On the one hand, we are helping customers to learn from Oracle Migrate to a lower cost database system . So I have always had a very special feeling about the domestic database , This is an emotion of love and hate . So today's final use “ chaos ” When it comes to this topic , Still a little hesitant , When the development of domestic databases is in full swing , Is it appropriate to pour this basin of cold water .

According to the white paper on database development of the Ministry of industry and information technology 2021 Description of , end 2021 year 6 End of month , Domestic relational database manufacturers alone have reached as high as 81 home , It is estimated that it will be released soon 2022 Breakthrough in version 100 Home even 150 Home is possible . Compared with a few more than ten years ago , In recent years, the development of domestic database industry is indeed very rapid , It is not too much to describe it as barbaric growth . Among these emerging domestic database manufacturers , There are also quite powerful genes , Huge investment , Enterprises that are really serious about the database industry , But there will certainly be mud and sand under the flood . It started late , Talent reserve 、 Domestic database industry with insufficient capital investment , Then it was cut into so many small units , The true ability of each individual is in doubt . Whether it's CPU, The server , operating system , Middleware these IT infrastructure , Devote oneself to IT The crown enterprises in infrastructure did not expect so many , I don't know whether it is the luck of China's database or the disaster of China's database .

Last Friday, I shared some speeches on the selection of domestic databases based on business scenarios at a salon , It is not to discuss business scenarios and database selection directly , It starts with the analysis of domestic database manufacturers . These analyses are based on the contents of the white paper on Industrial Development issued by the Ministry of industry and information technology .

In terms of years of establishment , Our domestic database enterprise is still very young , But true 20 There are still 14 enterprises with more than years of experience , It's just that these enterprises 20 It's not a good year , Taking database product sales as the main business cannot survive at all . So although there are 20 Years of history , In fact, the real history may have to be discounted .

Just looking at history may not be able to directly feel the gap , In terms of the number of employees , You can see the evil consequences of the fragmentation of the domestic database industry . exceed 60% Insufficient database vendors 100 people , More than 500 The human enterprise is insufficient 10%. For want to pick IT The Chinese Database Enterprise with the crown of infrastructure , The scale of the largest enterprise may not be as large as that of a small R & D Department of a second - and third rate foreign database manufacturer . If the personnel is subdivided into management 、 Research and development 、 product 、 market 、 sales 、 Logistics and other departments , I'm afraid there are even fewer R & D personnel . It is said that the R & D personnel of the largest database manufacturer in China are insufficient 500 people , This is the current situation of Chinese database enterprises .

If we look at the technical level , Judging from the number of patents ,90% The number of patents in the database field of database enterprises is less than 100 Pieces of , The total number of patents of all relational database manufacturers is insufficient 4000 Pieces of , The cut-off 2020 year ,Oracle The number of patents of a company exceeds 1 ten thousand 4 Thousand pieces .

The technical foundation is weak , Lack of talent , Why can so many database enterprises and products emerge all at once ？ From the technical source analysis of domestic databases, we can see some clues .

The chart above is made by us according to the collected data , Not necessarily very accurate , However, it can generally reflect the technical source of domestic databases . Most of them come from open source projects . Therefore, there will be a large number of smaller database enterprises . Using open source technology to develop your own database industry is not a bad thing , In fact, I am in favor of it . Making full use of open source technology can accelerate the development of domestic database industry , Shorten the gap with foreign leading enterprises . However, using open source technology does not mean relying entirely on open source technology , Instead, a lot of independent innovation should be carried out on the basis of open source technology , Add your own technology . For example, there are many domestic uses PG Open source database products , Which company is interested in PG The level of understanding of the source code , Yes PG The contribution of the community can reach Russia POSTGRESQLPRO And the level of ？ Gratifying is that , Behind this chaos, we have seen that some database manufacturers have started independent innovation , It has gone a long way on the basis of open source code , I want to have a few more years of accumulation , It will certainly break through some bottlenecks in open source technology , Walk out of your own path of autonomy .

The code localization rate of the database has always been a mystery , If you look at the code localization test report of the Ministry of industry and information technology , Then the vast majority of the database products that claim to be self-developed in China can come up with reports with a high rate of autonomy , And it's always 95% The above . I've tested a program that claims that the code localization rate exceeds 95% Database products , Its SQL The engine is completely “ compatible ”MYSQL Of , The storage engine does not use INNODB. Once I accidentally put a less commonly used MYSQL The original ecological parameters have been adjusted , little does one think ,SQL The working mode of the engine has been adjusted according to the requirements of the parameters . If only to keep MYSQL Syntax compatibility of autonomic code , Even this subtlety is so perfectly imitated , That's awesome .

Although there are few patents for domestic databases , However, this does not affect the domestic database to overtake on the curve , If you can't put Oracle I'm embarrassed to say that I am a domestic database . But the real application scenario reflects that our domestic database is in CBO and SQL Engine aspect and Oracle There is a big gap . I have also had in-depth exchanges with some database developers , They also admit that , To shorten and Oracle The gap is very difficult . No matter in talent accumulation 、 There is a huge gap between capital investment and feedback of practical application cases . Especially the third point , Lead to Oracle Optimizer problems can be found and improved from the production environment , And we don't even know what the optimizer is aiming for , Let alone planning the development route of the optimizer from the perspective of architecture .

Although our domestic database can not solve the urgent needs of users SQL Engine and optimizer improvements , However, it does not prevent us from innovating in other fields . On various publicity materials ,HTAP It has become the standard configuration function of domestic databases , But I don't think many domestic database manufacturers really know what it is HTAP. What they call HTAP Most are just one OLTP The database has certain batch processing capability .OLAP Computing scenarios and OLTP It's completely different ,OLTP Require a balanced distribution of resources , The delay of each execution is stable and as long as possible . and OLAP It requires large-scale or even giant computing under small concurrency , Use parallel execution to make full use of server resources , As much as possible CPU/ Memory /IO The ability to fully squeeze out , Complete complex calculations , High throughput data input and output . A database product that can't even isolate resources , How to support HTAP The two calculation scenarios that will hurt each other ？ Most of the things I've seen claim to be perfect solutions HTAP Problem database products , In fact, there is no practical mixed load capability . It's just that you can show some ability in some test environments .

Even so , It doesn't stop HTAP It has become a parameter index in the bidding of many enterprises , I don't know if the purchasing unit really needs this kind of computing power , Or is it just a gimmick to show partisanship against dissent ？

The last mess is the mess of the evaluation system , Every year, various so-called domestic database rankings will be published , However, this kind of ranking seems to be a bit like eating fruit in rows , The first and twentieth places are rated no more than 5 branch , I saw a list the other day , The difference between the first place and the tenth place is only 1 More points . If I were an enterprise IT executive director , There will be a feeling , The products on this list , It won't make much difference if you choose it casually . Ink sky wheel has a domestic database popularity ranking , Imitation DBENGINE Of , It is a relatively complete normalized list at present . However, the introduction materials about the database on the ink sky wheel are only the publicity materials provided by various database manufacturers , There is no relatively objective third-party evaluation . All the evaluation systems and rankings are good guys , This does not take advantage of the development of domestic databases .

Domestic databases now have the best development opportunities , We have seen the chip , The server 、 Security and other fields have shown great vitality when this opportunity comes . In the field of database that I am relatively familiar with , What I see is only a superficial prosperity , We have not seen a benign development trend , I hope this situation will change soon , I hope that the domestic database industry can rise suddenly .