Recently the students talk about the group more domestic basic software, online " database from Huawei to Ali database in the world, per mu jin a grandiose Stopped ," the article also made many forward. I am familiar with a number of databases, mainly involved in the last eight years GBase 8a MPP analytic database-related research and development, will share to you after finishing something.
1, based on the open source database developers with database product presence or absence of legal barriers?
First come the Baidu Encyclopedia https://baike.baidu.com/item/ open source license picture.
Figure 1 common open source license
Clearly, as long as the software company to follow the appropriate open source license, to comply with the law and the corresponding moral. For domestic software vendors, open source code certainly do not want a commercial product, it is mostly quoted BSD, MIT, Apache protocol, are closed source can be secondary development and publishing, if it is based on the GPL / LGPL / Mozilla open-source agreement they often have a corresponding commercial agreements, as long as the appropriate fees to the respective companies may also be used for commercial use.
As forced to amend the agreement, impostor, less common in basic software development engineers from speaking itself, is generally happy to follow the corresponding agreement.
Incidentally, in the field of database, Mysql, Mariadb series is the GPL (GPL infectious agreement), Postgresql series is the BSD license, which is Apache GreenPlum agreement. So the basis of domestic database to Postgresql family of software-based, distributed database to which GreenPlum based.
Answer why domestic databases are often on the grounds of the open-source database development, the most important of course, is the cost, you can stand on the shoulders of giants, but also have to start from scratch, then, to be profitable for the company is unrealistic. Also last 10 years, the database itself can be said that there is no revolutionary / disruptive new theory, the emergence of new algorithms, (Stonebraker create Vertica 2005) by my former colleague Mengxiang Bin argument, the last revolutionary change or OLAP ( analytical) data warehouse technology business in rough set index columns exist in the application database, thus the birth of Sybase IQ, Vertica and a series of OLAP products.
I personally think that from 2000 onwards, is driving the development of a database client application, whether it is a column store database, map database, full-text database, KV Hadoop database or the many eco-products, in addition to Oracle database Yiqijuechen in matters other products because certain properties to meet certain business applications, resulting in market segmentation to develop on either Ali or South common, database vendors are taking Oracle "hanging drop", one that is XXX times faster than Oracle, in fact, Oracle should not have suited the business scene.
As the data on billions of rows of a table to add, delete one, Oracle may take days, and the column store database do not have even one second, this comparison would not make much sense, but the actual business but very common, the fundamental reason is that the business scenario should simply did not use Oracle, as Stonebraker said single model can not conquer the world.
Correspondingly, "integration", "mix and match", "integration" does have technical feasibility, will open up each database integration, facilitate the development, operation and maintenance and lower overall operating costs of programs and products in recent years, more than a color of the Department .
2, similar problems, based on open source database developers obstacles it on moral or without the database?
Daily life, so that everyone is sick of the country based on open source software development, OS, database software is open source with the name of "completely independent intellectual property rights" in the name of illegal occupation of state funding as well as over-marketing cases.
" Illegal occupation of state funding " is the scope of those responsible for the approval of leaders and experts concern, of course, be punished by law, " over-marketing " really deceived us feelings of the public, which are those that can lower the customer's problem, like someone non buy "Melatonin" the same, but not afraid, low-level customers to choose wrong, and certainly by the superior end user or punishment, critical infrastructure software, especially database software, this ratio is still very low, and then add the following reason.
We can talk about the "independent intellectual property rights" slogan. Personally, I think this is the rise of someone with great leadership / big expert, think of a word, he has no legal and moral meaning, like "Huan crossing the Spring Festival." In fact even if you write on Linux a "Hello World!",
#include <stdio.h>
int main(void)
{
printf("Hello World!\n");
return 0 ;
}
To be used in the following link library
Figure 2 Hello World is dependent libraries
So in theory, we should declare the copyright notice, this software uses software libraries such as glibc such as the GPL. The same is true on the Windows platform, even if all the libraries can statically linked, what changes do not occur in nature.
I believe countries also understand the "completely independent intellectual property rights" is a false proposition, now the official statement said less, but began to promote " self-control ", " safe and reliable ", that the recent amount of up to hundreds of billions of scale AK (encore) project, and later found a similar national Ali joint venture or listed companies can not participate, after another "actual controller" argument.
3, case analysis
Case: South Common in 2015 by the Ministry of Industry (People's Republic of China Industry and Information Technology) brokered the purchase of the source code for the Informix database from IBM to spend huge sums of money (IBM sold a total of 3 parts source in China), we were Cui ( Cui Wei force ) and a leader Ministry confirmed after Informix source code reform GBase 8t relational database (later renamed GBase 8s) count as the country's " self-controlled ", " safe and reliable " products, the leadership made it clear that "count" in this project then we use products to participate in national security, military and other fields prove indeed counted as AK software.
By extension, based on open source code and claims to be able to fully grasp: to modify the bug, enhanced security (in line with state secret XXX), can be optimized to provide technical support and custom development, also in line with the law (not violate open source licenses, software copyright registration and software patents) of database software, can apply for the AK database.
4, soft paper on aspects of the database
Soft paper often appear in some IT news sites, from the media, and even the software company's official website, it is a mode of marketing, like comic racing season "bragging", with a positive blowing, the opposite of blowing, have belittle others blowing, blowing have to belittle their own, we see a lively enough.
The core is our definition of software engineering, the core of the software is to bring value to customers / users, and now the domestic infrastructure software is very difficult to completely a buyer's market, the project POC test (Proof of Concept, user scenarios verification testing) stage is very long we often need to do a project to deploy a couple of years, it is the customer / user will have all their APP to run for a long time, until all of the functionality and performance to meet the demand, and also made a lot of customization needs, and then some products are measured round or rounds, will start the bidding, the bid if lucky, you sign the agreement, deploy, and run a very long time before the opportunity will give you the money (no riches, which is the capital market one of the important reasons to abandon database infrastructure software), distributed database project often appears company a hard two or three years in a project, all the transplant, optimization POC are done, then grab one of company B phenomenon.
So at least for now, the real customers will not impress these soft paper, of course, a tragic manifestation of which is the domestic infrastructure software competition. We unlike the United States, DOD has spent large sums of money training and management software market, we are a large number of tertiary hand to the company, last year to participate in a XX system project, involving dozens of companies large projects, only the South Common, a non-XX the tertiary industry, the state invested heavily in IT, actual results will need time to verify. The country's large investment, including AK, because provinces and cities to encourage local companies, so the past two years there have been dozens of new database company (other estimates for sub-areas of the newly formed AK software companies a lot of it), AK the end result, we can only say wait and see.
Infrastructure software vendors soft paper often not to the public to see, is to their co-workers to see themselves encourage yourself, the recent South Common soft paper it is also planned to welcome the students throwing bricks.
5, regarding the evaluation database
Whether it is a test of the Ministry of Industry Ali was similar to the TPCC to participate in the tests are marketing tools to promote the domestic software market with the bidding of various companies are constantly pushing finalists threshold. In recent years, many countries have such institutions, to participate in a common database of South China Institute of Information and Communication (ICT referred to the hospital) up to the test. As for how the quality, we test during the reference application software copyright on it, take seriously the process has been the performance of the best.
Foreign TPC series of tests (TPC-C for transactional database, TPC-H, TPC-DS database for analysis, as well as test for the Hadoop platform), they are characterized by openness is better, the standard contains data generation , assessment tools, the selected scene is typical, but also evolving, such as TPC-H is not representative gradually after launch more advanced TPC-DS. In fact, two years before the star ring technology also vigorously promote the results of the first TPC-DS, it may not pay attention to nothing. Reference: CCTV channel three columns centralized authority reported that the star ring technology by the world's first TPC-DS test
The second is the ranking of the various domestic CCID, ICT Academy represented, can only say that ranking is relatively fair, in recent years for more money to add a lot of dimension data processing industry, resulting in the division of fine enough to enter the rankings many companies, gives the impression of gold declined.
From abroad is big for Gartner authoritative better, its own research on technology trends are also a little more, below is the 2018 and 2019 Gartner released data management direction of the Magic Quadrant (Note: Gartner and enter into the Gartner Magic quadrant are two things that way, Sequoia has also been found to promote their inclusion Gartner does, in fact, Gartner's list, and indeed very great). China is entering the South Common, Huawei, Ali (Ali and Huawei which points more in terms of cloud). Published in 2018 reported that domestic database vendors have been star rings database, the database giant sequoias, two years gone. This selection of product strength is important, but the selection itself costs money and effort is very small or less-specialized database is no way to participation.
Figure 3 Gartner2018年初发布的报告
In the 2018 report, Ali (Alibaba Cloud) and Huawei Technologies (Huawei) of the body than the South Common (GBase), technological advancement is better GBase. In the 2019 report, Huawei surpassed South Common. One reason is that Mr. Cui Wei forces died unexpectedly, South and GM products are not stable development in 2018.
Also graphics in these two years, we can see ahead of Oracle, domestic database also did not enter the Leaders Quadrant, we still need to work that way.
FIG 4 Gartner2019年初发布的报告
Secondly, there is a database ranking https://db-engines.com , authoritative worse, for reference it.
summary
1, or the use of open source database to the open source database developed based on its own database, as long as the act according to the agreement, on legal compliance, have nothing to be ashamed of.
2, cheat national funding evade the law, we lie to the feelings of the database software is still relatively small.
3, database software is a buyer's market, POC testing is an important cost, customers generally will not because the manufacturers of test marketing on relief.
4, the network soft paper is to give everyone a happy, not too serious.
5, the state vigorously promote AK software, specific effects need time to prove.
6, the quality of domestic evaluation and ranking general, good quality of some foreign Gartner, the threshold is too high, but these direct benefits in bidding for customers reference only.
7, "a short step, a thousand miles; not small streams into a mighty torrent", Ali's progress this is a solid step forward domestic infrastructure software, if its some boast, let these values progress to boast of some more of it.
The next article and then explain migration path to South Common GBase 8a MPP analytic databases, as well as some of our tangled and open source software. Thank you for taking the time to see this short article.