Why does NewSQL eclipse the traditional relational database (try to explain the difference between newsql database and traditional relational database)

Traditional relational databases may never disappear-at least not soon, but their glorious days are gone forever.

The popularity of many emerging NoSQL databases, such as MongnDB and Cassandra.

This makes up for the limitations of the traditional database system.

Compared with the vigorous development of NoSQL, the relational database system based on SQL does seem a bit lifeless.

But this is the fault of the database vendor, not SQL.

Relational database has long been a key part of enterprise deployment, but now it has a better choice to adapt to the new data structure and modern hardware system.

Suppliers such as IBM, Microsoft and Oracle Bone Inscriptions will continue to use relational databases to dominate the core functions of their financial transactions.

But NoSQL database seems to be more suitable for today's era of massive data.

Such as ApacheHadoop and MapRece technologies.

RobinBloor, chief analyst of Bloor Group, said that the traditional relational database is outdated and its architecture needs to be updated.

Bloor's reason is that with the continuous maturity of multi-CPU computer and solid-state hard disk technology, it is no longer important to access the data on the disk.

Solid state drives are faster, so the reading rate between disk and memory will increase.

DanMcCreary, a data strategy consultant in Minneapolis, Minnesota, pointed out that SQL database has its own problems, such as poor scalability.

When the data growth exceeds the limit that one server can bear, it is necessary to share or divide the data to multiple servers, which is a complicated process.

In addition, such as problems caused by external links.

For example, data fusion of multiple tables, performing some operations across servers may cause some problems.

The rise of NoSQL and the appearance of "NewSQL" will change the definition scope of data.

It is no longer the original data type, such as integer and floating point.

The data can be an entire file.

NoSQL may scare database administrators because they are worried about losing their domain names.

NoSQL database is non-relational, horizontally extensible, distributed and open source.

DwightMerriman, founder of MongoDB, said that NoSQL can be used as the background storage of Web application server, content manager, structured event log, server side of mobile application and file storage.

MichaelStonebraker, chief technology officer of distributed database company VoltDB, said that NoSQL databases can provide good scalability and flexibility, but they also have their own shortcomings.

Because SQL is not used, NoSQL database system does not have the characteristics of highly structured query.

Other problems with NoSQL include the inability to provide acid operations (atomicity, consistency, isolation and persistence).

In addition, different NoSQL databases have their own query languages, which makes it difficult to standardize the application interface.

Stonebraker said that the lag of database system can usually be attributed to many factors.

Such as a buffer pool maintained by a database system for recovering logs and data fields for managing locks and locks.

In the test of VoltDB, it is found that these behaviors consume 96% of the system resources.

The data processed by RDBMS is only about 16%. "Although relational databases feel the pressure of new technologies, RDBMS still occupies some positions in enterprise computing.

At present, the market of RDBMS is about $35 billion.

These include software licensing, services, technical support and account maintenance, "said NoelYuhanna, an analyst at Frest.

Forrester predicts that 25% of business data in enterprises will be structured data, of which at least 65% are using RDBMS or other traditional relational databases, and RDBMS accounts for at least 65,438+06% of transaction data.

Enterprises will have 75% business data combined with semi-structured files (such as XML, e-mail and EDI) and unstructured data (such as documents, pictures, audio and video).

Yuhanna said that about 5% of the data resides in relational databases, and the rest is distributed in non-relational databases and file formats.

In addition, I am worried that column data will become an excessive candidate for changes in the database field, and they may make relational database products simpler.

Traditional relational data vendors such as IBM, Microsoft and Oracle must have new plans in their RDBMS field, and they will not choose to disclose their plans.

Bloor said that no one would notice that RDBMS might die.