Kunlun distributed database technology advantages

The last chapter introduces the technical characteristics of Kunlun distributed database , This chapter follows up on the technical advantages of databases .

One 、 Open source 、 Raise the good and avoid the bad

1.1  Set 3 Major databases Oracle、MySQL、PostgreSQL All of our strengths come into being 1+1>>2 Amplification effect of , Community energy can be widely gathered , Produce long-term value and vitality .

• Oracle: Storage engine , Query processing .

• innodb Fully comply with Oracle The design of storage engine .

• MySQL:innodb Storage engine and binlog Copy (RBR).

• PostgreSQL: Query processing capabilities in all open source RDBMS The strongest .

• MySQL&PostgreSQL Human and technical resources of the open source community .

1.2  Avoid their weaknesses

• Oracle: Hardware and software are expensive , High cost : Unable to be safe and controllable , Policy compliance .

• MySQL: Query processing has limited performance and functionality : Stand alone database , Unable to expand the horizontal elastic capacity .

• PostgreSQL: The storage engine is not suitable for heavy loads OLTP load .

Two 、 Query processing capability

2.1  Computing nodes support PostgreSQL All the main query processing functions of .

• Support the vast majority of DDL And all DML Grammar and function .

• exception : Foreign keys and triggers ,tablespace And storage related functions ,WAL replication.

• All basic data types are supported .

• The number , character string ,text/blob, Time 1 date / Time stamp /money/enum, Sequence and so on .

• Support advanced query processing functions .

• Any span shard Multi table join of , Subquery , stored procedure .

• OLAP Analytical ability : Aggregation function ,window function ,grouping sets, cube, rollup.

• CTE, View , Materialized view , really prepared stmt,jit.

2.2  Compute node compatibility MySQL and Oracle Common use of SQL grammar (*).

• Support MySQL Client protocol (*)

• Go to O The migration workload is small , Technicians' skills can be translated , The learning curve is smooth (*)

3、 ... and 、 Comprehensive data security

3.1  Controlling data access at the data source is more secure and reliable .

• Unified / Multilevel / Flexible and dynamic configuration of access control rules .

• Many defects of application level access control .

• No unified : Multiple applications access the same database , Every application needs rule configuration and even coding implementation .

• Inflexible dynamic : Hard coded access control rules , It's not easy to modify .

• unsafe : Control policies and rules themselves can reveal information .

3.2  Multiple levels of fine-grained access control .

• Multi tier users / role .

• Multi tier database objects : database /schema/ surface / View / Column .

• Access control rules for managing various database objects at multiple levels .

3.3  System wide encryption of user data : Data files and binlog Files and log files can be encrypted ;

SSL Full link encryption , Secure data transfer ;  multistage （database,schema,table,view,column） Fine grained access control .

Four 、 fully inclusive and equitable

4.1  Computing node open architecture .

• Extension: Seamless compatibility PostgreSQL ecology ,PostGIS etc. .

• FDW(foreign data wrapper): Interfaces can be implemented to read all major data sources .

• hadoop ecology :hbase,hive etc. .

• Mainstream databases :Oracle,MSSQL Server,DB2,MySQL,PostgreSQL wait

• Column store (OLAP) And temporal databases .

4.2  Other advantages of computing nodes .

• Perfect i18n/globalization/localization Support .

• The time zone , Character set and collation.

• Multilingualism .

5、 ... and 、 Multi tier and multi-faceted scalability

5.1  On demand elastic horizontal expansion capability .

• Multiple read / write nodes , Processing both read and write loads can expand processing capacity on demand .

• No sharing (sharenothing), No single point of dependency .

• The bottleneck of asexuality , No calculation / Storage capacity bottlenecks .

• Add or remove compute nodes and storage clusters as needed / node .

• Transparent on-demand expansion , Business systems and end users are not aware .

• The storage cluster expansion speed is adjustable , Calculation of data source nodes / Storage / The consumption of network resources is controllable .

5.2  System wide parallel computing capability .

• Give full play to the parallel working ability of the server .

• Multicore parallelism .

• Storage system parallelism .

• Network system parallelism .

6、 ... and 、 Other advantages

6.1  No fixed number of partitions , Instead, you can have a different number of partitions per table , Zoning rules （3 Kind of ）, And you can use any number of partition Columns , Even small tables can be completely UN partitioned .

6.2  Storage cluster

• Performance Leadership : The distributed transaction processing performance is much higher than that of the community version http//www.zettadbcom/blogs/perf-cmp-mysql.

• Complete disaster recovery capability : Fill in the community version MySQL80 The disaster tolerance capability of distributed transaction processing is blank .

• https://fosdemorg/2021/schedule/event/mysqlxal

• The domestic address of this technology sharing video :https//wwwbilibilicom/video/BV1zo4y1d7pu

• Native onlineDDL function : Quick addition .

6.3  Kunlun database VS MySQL: Advantages of using Kunlun database to manage small-scale data .

• Zoom in on the single -MySQL The ability to cluster , On demand horizontal expansion capability and stronger data analysis capability .

• Easier to use MySQL colony : Automatic switching and maintenance mysql State of the cluster .

• Parallel query processing , Standby machine reading .

6.4  The cluster structure is simple , Independent of third-party modules and software (etcd/zookeeper etc. )

• Product quality is controllable .

• The labor burden is small .

Be careful ： upper （*） The entry for is the feature to be supported