Shulou(Shulou.com)06/01 Report--

Introduction

Someone left a message on an official account that focuses on SQL as follows:

Contempt for SQL Server

This message touches on a very sensitive question: is there a future for relational databases? Now it is 2020, the blockchain is hot, AI talent has been "surplus", big data has become a common radish cabbage, but you are still working on SQL Server? Are you still working on SQL?

Reason

In the last 10 years, the changes in the field of data management have been brilliant. Hadoop, Spark,MemcacheDB, Redis, Vertica, HANA, Cassandra, MongoDB, InfluxDB,TiDB and so on, to name a few. It is true that all kinds of high-end summits and forums only see AI and big data, and there are few traditional relational databases.

There are four reasons for the preliminary analysis:

One of them

The rise of cloud computing and the large-scale application of open source databases have greatly reduced the cost of computing power and data storage.

Take Oracle, the leader of traditional relational data, as an example, with the rise of the Internet, a large number of users have generated huge amounts of data, but when Oracle is deployed on a large scale, not only the Oracle software itself is expensive, but also the storage expenses of IBM minicomputer + EMC are also huge. For example, moving an Oracle data cluster from one room to another in the same data center costs more than one hundred thousand yuan. The maintenance cost of software and hardware, including the requirements for administrators, is also very high.

In addition, Oracle itself has an upper limit on performance, followed by a closed source black box. For unencountered scenarios, no matter how hard you try, it is impossible to predict the problems that may arise.

Most of China's Internet companies are reckless and attach great importance to performance-to-price ratio. The then Internet leader, Yahoo of the United States, began to take the lead in using MySQL databases, and Chinese Internet companies followed suit.

The most typical example is Alibaba. Alibaba was once known as Oracle's "star client" and "Whampoa military Academy". However, with the surge in the number of registered users, there is more and more data generated by users. Taobao has enabled the largest Oracle RAC cluster in Asia. The amount of data of Alibaba B2B Chinese station is also due to large data volume and business requirements. Between 08:00 and 09:30 every year, the utilization rate of CPU is 98%, and the server load is also extremely high. Even using hot and cold isolation can not solve the problem of large capacity of data and large concurrency. This will happen again soon after the storage device is replaced.

Alibaba was forced to "save" a technical route to IOE: "low-cost, linearly controllable, decentralized (that is, distributed): replacing minicomputer with IBM,PC Sever, replacing minicomputer with Oracle, replacing with MySQL, removing EMC and using medium-and low-end storage."

In the subsequent explosive growth, including the business peak represented by Singles Day holiday in 2014, MySQL branch based on the open source version of Ali well supported the extremely stringent business requirements, and finally the MySQL, which is very convenient to cloud and open source at the same time, gradually began to replace the quack status of Oracle in domestic Internet companies and became the first choice for many Chinese Internet companies.

Second

Reduction in the cost of general hardware. With the effect of Moore's Law, there are even many in-memory or distributed in-memory databases such as HANA, Vertica,InfluxDB.

So how cheap are cloud computing and cloud storage now? Take a look at the ad you just intercepted from a website today:

The advertisement of the CVM on a website

This kind of product that is sold on a large scale to the market, because it can use cheap PC Server, can achieve the real price of cabbage, which would have been unthinkable 10 years ago.

Third

In addition, the MySQL code and documentation are public, so the requirements for database developers and maintenance personnel are greatly reduced.

Last

From SQL to NoSQL, databases that support semi-structured data and unstructured data shine brilliantly. Previously, this kind of unstructured data problem has been solved, and unstructured data processing has become the standard configuration, which seems to occupy the hot spot of the market. However, NewSQL is still in development.

Develop

The development direction of the database roughly includes the following aspects:

-from stand-alone and cluster to completely distributed. At present, databases such as Oracle have been implemented.

-the integration of OLTP and OLAP, which has been implemented for a long time.

-structured and unstructured fusion, which has also been implemented by Oracle, MS SQL Server, etc.

-the integration of disk and in-memory database has also been implemented by Oracle, MS SQL Server, etc.

-the fusion of different types of databases, such as rows, columns, graphs, objects, documents, content, timing, search engines, navigation classes, etc., the author finds that a small part of them have been implemented.

-at first glance, the support of data modeling tools for the above features is also being followed up, but it is still a long way from perfect support.

To sum up, in the future, vertically, database individuals think that the storage of pictures, audio and video will continue to become a hot spot, and horizontally will further integrate and develop, such as the integration of audio and video storage and structured data, and so on.

Conclusion

So is the traditional relational database a thing of the past? If you think so, you are making a big mistake! Let's take a look at the authoritative data rankings:

Trend of data popularity in recent years

Database popularity ranking in December 2019

It can be seen that although unstructured data is hot, the relational database is still sitting in Diaoyutai!

What is the reason? The development of the database is colorful, but in the end, the key lies in the business scenarios it is good at. Different types of databases adapt to different business scenarios. There are a large number of structured data in the real world, relational databases deal with them relatively mature, after decades of development and a variety of harsh tests, all aspects of advantages are unique. Now big data of the fire focuses on effectively solving the age-old problem of semi-structured and unstructured data, which has been unable to deal with before. To let an unstructured database play an important role as a relational database is tantamount to letting the excavator race with the big truck to pull the goods.

In the past, relational databases did occupy the absolute mainstream. However, with the development of 3G and 4G in recent years, the content of pictures, audio and video has increased slightly. With the development of new technologies such as 5G, AR\ VR, Internet of things and so on, storage will become more expensive, blockchain and quantum computing will also bring new changes to the world. But no matter how it develops, I think relational database will still be the mainstream in the next decade, and it will have an important position in the future. Because structured data will still be the most important digital asset for human beings in the foreseeable future.

Original quality assurance, follow me, share data-related practical information every week!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.