GBASE NTU General Technology Sharing:
1. Description of requirements.
The database system is the core component of the information system and the necessary condition for the normal operation of the system, so the security and reliability of the database system are required to be high, especially in the core business system. A centralized database is one that puts all databases in a central point, and all businesses are connected to this database for read and write operations. This model has the advantages of simple management and O&M, but there are hidden dangers of single points of failure and performance bottlenecks, and the sustainability of the system cannot be guaranteed. In order to solve these problems, enterprises generally use the method of two cities and three centers to deploy centralized databases to ensure business continuity.
The following is an example of the GBASE 8S centralized transactional database
Second, the introduction of the program.
The primary center deploys SSC clusters for load balancing, multiple nodes in the cluster provide read/write services, and the disaster recovery center in the same city deploys HAC nodes to ensure that the primary center is down, and the standby center is quickly switched. The remote disaster recovery center deploys RHAC nodes to achieve remote disaster recovery of thousands of kilometers, and can be forcibly converted to the primary center in the case of manual intervention, and can also support automatic switching to the primary center when the equipment in the primary center fails. Multi-center data centers are managed in a unified manner through the CM connection manager, and service requests are sent to specific cluster nodes based on SLA rules.
3. Characteristics of the program.
It has a hot standby function.
In the event of a system failure, a highly available database cluster can automatically switch over the standby server to restore the service capability of the database to ensure business continuity and minimize the negative perception of customers due to business interruption.
Data consistency between active and standby systems.
If the database switch is completed due to a fault at this time, the customer will continue to use the follow-up service after the switchover is completed, so as to avoid data inconsistencies caused by the switchover and let the customer start over again.
The switching process is done automatically.
The primary/standby switchover process in the cluster is automatically completed by the database management system without manual interventionThe business system is not obviously perceived during the switchover process, and the business system does not need to be modified in any way. For example, if the business is interrupted due to database software and hardware failures, the business system can automatically resume normal operation after the database is restored.
Switching time in seconds.
Ensure real-time synchronization of standby and host data, consistent active and standby data, and support fault detection and standby automatic takeover of host services, RTO<20s, RPO=0
Long-distance disaster recovery capability.
When all database nodes in the production computer room lose data service capabilities due to power outages, equipment damage, and other reasons, the high-availability database cluster can have long-distance disaster recovery and switching capabilities of 100 or thousands of kilometers to ensure business continuity.