Klustron technical advantages

Klustron is used in various industries to store, manage and utilize massive relational data, supports high-concurrency and high-load transaction processing and data reading and writing requirements, can automatically complete horizontal elastic scaling without stopping the server, and ensures continuous data retention in case of node and network failures The continuity of read and write services, as well as data persistence and consistency, while maintaining a stable high throughput and low latency. Klustron is compatible with MySQL and PostgreSQL connection protocols and DML SQL syntax, and applications developed in all common programming languages can be seamlessly connected.

01 Klustron has the following series of very unique and distinct technical advantages.

1.1 Indestructible

Klustron can automatically respond to cluster environment failures --- any node failure or network failure of a Klustron cluster will not lose or damage user data, nor will it affect the normal operation of the business system. Klustron's storage cluster provides financial-level data consistency and high availability based on self-developed fullsync technology. After the failure of the master node of the Klustron-storage storage cluster, the Klustron cluster can quickly detect the failure, then automatically select a new master node and complete the master-standby switchover, so as to continuously provide database services. Klustron can ensure that the failure of any node in the cluster at any time will not lead to loss or confusion or damage of user data, and continue to provide data management services.

1.2 Elastic scaling

Klustron supports non-stop horizontal elastic scaling, allowing users to flexibly plan, allocate, and use computing and storage resources according to dynamically changing business loads. Auto scaling will not have any impact on business systems and logic, as well as end-user experience. The Klustron cluster automatically completes horizontal elastic expansion and contraction, utilizes all servers in a balanced manner to store and manage user data, and provides continuous and stable high-performance and low-latency OLTP data access capabilities.

1.3 Inclusive of all rivers

Klustron is the only one in the industry that supports the connection protocol and SQL syntax of both MySQL and PostgreSQL databases, and can seamlessly connect MySQL and PostgreSQL, the two world's top open source database ecosystems. The client connection library of all common programming languages ensures that software written in these languages can connect to Klustron and correctly execute all standard SQL statements, as well as MySQL and PostgreSQL's private DML SQL statements. The corresponding data migration work can also be completed with one click with the help of Klustron and third-party data import tools.

1.4 Extreme Performance

The OLTP performance of Klustron is far ahead of competing products, ensuring that terabytes of data, thousands of connections, and hundreds of thousands of QPS can still provide transaction processing performance with high throughput and low latency. See performance comparison and test report for details .

1.5 Ultimate Security

Klustron supports full data encrypted storage, log encrypted storage, and full link encrypted transmission; at the same time, Klustron supports multi-level fine-grained access control, so that DBAs can define access control rules in the database system, so as to ensure that user data will not be used by any application Software unauthorized access.

1.6 User Friendly

Klustron provides a series of DBA common tool sets (GUI and API call interface), which is convenient for DBA to complete cluster management with one click or programming (cluster creation, node startup and shutdown, addition and deletion of computing/storage nodes, addition and deletion of storage clusters), redo standby, Active/standby switching, physical/logical backup recovery, online DDL, elastic expansion and contraction, repartition, unified retrieval of cluster node log files and operation status monitoring, etc., all complex tasks that DBAs need to handle greatly improve the work efficiency of DBAs and avoid DBAs manual work. Operations lead to unexpected errors that affect the operation of business systems.

02 Detailed explanation of product advantages

The following is a detailed interpretation of the above-mentioned advantages of Klustron.

2.1 Foster strengths and avoid weaknesses, 1+1>>2

Klustron takes advantage of the strengths of the two stand-alone databases, MySQL and PostgreSQL, and avoids their respective weaknesses, achieving the effect of 1+1 >> 2. Specifically, when the PostgreSQL database handles high concurrent OLTP loads due to the inherent defects of the storage engine, it cannot achieve continuous and stable high-performance data updates, and this is exactly the strength of MySQL --- MySQL innodb fully complies with Oracle's storage engine. Design, with excellent OLTP transaction processing performance, especially its performance advantage is more obvious under large concurrent heavy load; at the same time, MySQL has poor performance and limited functions in table connection and OLAP analysis query, while PostgreSQL has excellent performance in read-only query And the function completely covers all the functions of OLAP.

The Klustron team formed the computing node Klustron-server of Klustron based on the in-depth research and development of the PostgreSQL kernel, and formed the storage node Klustron-storage of Klustron through in-depth research and development of MySQL. Several computing nodes and storage nodes form a Klustron cluster, thus realizing OLTP transaction performance far exceeding PostgreSQL And far beyond the query processing and OLAP analysis capabilities of MySQL.

Both MySQL and PostgreSQL are stand-alone databases. The amount of data they can effectively manage is limited by the computing and storage resources of a single server. In practice, it is difficult to provide continuous, stable, high-performance, low-latency data management capabilities that are acceptable to the business. .

It can be seen that Klustron has a qualitative leap compared to PostgreSQL in terms of scalability and performance --- the measured performance data also supports this conclusion --- it is the best choice for OLTP databases using PostgreSQL application software and information systems .

At the same time, users use Klustron in exactly the same way as using stand-alone MySQL or PostgreSQL, without any special application-side development work, so the existing skills of application programmers and MySQL DBAs can be translated to Klustron, and the learning curve is very flat. In general, Klustron is rooted in the two world-class open source communities of PostgreSQL and MySQL, and the available human and technical resources are huge.

Klustron has been connected to Alibaba Cloud and AWS, and will soon be connected to other public cloud platforms to provide users with distributed relational database services (DRDS) and pay-as-you-go DBaaS services.

2.2 Query Processing Capabilities

2.2.1 The computing node supports all the main query processing functions of PostgreSQL

For the compatibility of Klustron with PostgreSQL, please refer to this article .

• Supports most DDL and all DML syntax and functions

  • Exceptions: foreign keys, tablespace and storage related functions, WAL replication

• Supports all primitive data types

  • Number, string, text/blob, time, date/timestamp/money/enum, sequence, etc.

• Supports advanced query processing functions

  • Any cross-shard multi-table join, subquery, stored procedure
  • OLAP analysis capability: aggregate function, window function, grouping sets, cube, rollup
  • CTE, view, materialized view, real prepared stmt, jit.

• Benefit from PostgreSQL's good compatibility with Oracle, the workload of migrating data and application systems from Oracle database to Klustron is relatively small

  • PostgreSQL supports most of Oracle's SQL syntax and PL/SQL syntax
  • Support third-party data migration tools (ogg, etc.)
  • Skills of technicians can be translated, and the learning curve is gentle

2.2.2 Computing nodes are compatible with MySQL DML function

• For details about Klustron's support for MySQL's private DML syntax, see Klustron's support for MySQL's private DML syntax
• For details about Klustron's MySQL connection protocol, see Klustron's MySQL connection protocol introduction
• For the subtle differences between Klustron and MySQL syntax, see MySQL syntax and function summary not supported by Klustron

2.3 Comprehensive Data Security Guarantee

2.3.1 Controlling data access at the source of data is more secure and reliable

• Unified/multi-level/flexible dynamic configuration of access control rules
  • Klustron allows users to control data access permissions at the database, schema, table, column, view, and even row levels
  • Access control rules should be configured in the database system rather than relying on the application to constrain itself
• Various flaws in application-level access control
  • Inconsistency: multiple applications access the same database, and each application requires rule configuration and even coding implementation
  • Inflexible and dynamic: hard-coded access control rules, not easy to modify
  • Insecure: Control policies and rules themselves can leak information

2.3.2 Multi-level fine-grained access control

• Multiple levels of users/roles
• Multi-level database objects: database/schema/table/view/column
• Multi-level management of access control rules for various database objects

2.3.3 Data Encryption

• System-wide encryption of user data, data files, binlog files, and log files can all be encrypted
• SSL full link encryption, secure data transmission
• Multi-level (database, schema, table, view, column) fine-grained access control

2.4 Open Compatibility

2.4.1 Computing Node Open Architecture

• FDW ( foreign data wrapper ): interface can be implemented to read all mainstream data sources

  • hadoop ecology: hbase, hive, etc.
  • Mainstream databases: Oracle, MSSQL Server, DB2, MySQL, PostgreSQL, etc.

• Perfect i18n/globalization/localization support

  • Time zone, charset and collation
  • multilingual ability

2.4.2 Compatible with MySQL binlog ecology

The binlog events output by the storage cluster can be streamed for consumption by third-party tools.

2.5 Multi-level and multi-faceted expansion capabilities

2.5.1 On-demand elastic horizontal expansion capability

• Multiple read-write nodes can handle read-write loads and can expand processing capacity on demand
• No sharing ( share nothing ), no single point of dependency
• Increase or decrease computing nodes and storage clusters/nodes as needed
  • No performance bottlenecks, no computing/storage capacity bottlenecks
• Transparent on-demand expansion, no perception of business systems and end users
• The expansion speed of the storage cluster is adjustable, and the consumption of computing/storage/network resources of the data source nodes is controllable

2.5.2 Parallel computing capability of the whole system, making full use of a large number of servers to work in parallel

• Multiple threads inside the computing node execute the same query in parallel
• Multiple query subtasks dispatched by computing nodes for the same query are executed in parallel by multiple threads in the storage node
• Computing nodes send requests to storage nodes asynchronously, allowing multiple storage nodes to execute multiple subtasks of the same query in parallel

2.6 Other advantages

2.6.1 Storage Cluster

• Leading performance: distributed transaction processing performance is much higher than the community version
• FullSync financial-level high availability and strong consistency, node and network failures will not lose or damage data
• Complete Disaster Recovery Capabilities: Fill the Disaster Tolerance Capabilities of Distributed Transaction Processing of Community Edition MySQL8.0

2.6.2 Klustron VS MySQL: Advantages of using Kunlun Database to manage small-scale data

• Enlarge the ability of single MySQL cluster, on-demand horizontal expansion ability and more powerful data analysis ability
• Easier and more convenient use of MySQL clusters: automatically switch masters and maintain the status of MySQL clusters
• Parallel query processing, read-write separation (prepared for machine-reading)

2.6.3 The cluster structure is simple and does not depend on third-party modules and software (etcd/zookeeper, etc.)

• Product quality controllable
• Avoid additional hardware overhead
• Small operation and maintenance burden

END