Geo-redundant disaster recovery solution

Two places: in the same city or in different places

Three centers: production center, same-city Dr Center, and remote Dr Center

Hypermetro data center: Two data centers are running. They run the same applications and have the same data. The two data centers provide cross-center service load balancing capabilities, enabling continuous application availability and disaster backup

Data Dr Technology selection metrics:

Disaster tolerance: It is clear that the computer system needs to withstand the type of disaster, system failure, communication failure, long-term power failure, fire, earthquake and other unexpected situations to take different backup, protection plans.

Business impact degree: It is necessary to define the loss degree caused by the business pause when the computer system is unable to work due to an accident, that is, to define the maximum tolerance time for the failure of the computer system. This is an important technical indicator for designing a backup plan.

Data protection degree: Whether the database is required to recover all the submitted transactions and real-time synchronization is required to ensure data continuity and consistency, which is an important basis for the complexity of the backup plan.

The features of the scheme are as follows:

  1. In WYSIWYG Dr, the standby system is directly online and can be directly visible and verified. Unlike other DISASTER recovery systems, you can’t know whether the backup system is good or bad until the recovery is complete.

  2. The application-level replication technology means that the data replicated by the mirror system is database transactions and belongs to the application layer, thus ensuring the integrity of database data.

  3. Implementation of business system without interruption, suitable for 7X24 hours of continuous operation of the business system.

  4. There is no need to ensure the consistency of the hardware of the active and standby systems, which greatly reduces the cost of system transformation and hardware investment. The backup storage space is only a little larger.

  5. It consumes very little network bandwidth and does not need dedicated optical fiber transmission network. The real-time incremental replication technology greatly reduces the resource overhead and has little impact on service system performance.

  6. Once the primary system fails, the database of the standby system is in the running state directly and only the service system can be restored without the data recovery phase. Therefore, the whole process of the standby system replacing the primary system is very fast.

  7. A web-based unified management platform for configuring servers and databases, setting mirroring policies, and monitoring the running status of replication chains. You can monitor backup jobs and running status of the entire network on the WEB monitoring page. Each monitoring record is a complete replication link, including mirror agent, mirror server, and running status.

1. Real-time data disaster recovery case

1) Data normal operation Data replication architecture diagram

2) If the production server is faulty, the Dr Server continues to support the fault

3) When the production system is restored

4) Dr System advantages

**1, ** Log analysis-based database Dr System advantages: ● What you see is what you get Dr, the standby system is directly online, is directly visible and verifiable. Unlike other DISASTER recovery systems, you can’t know whether the backup system is good or bad until the recovery is complete. Application level replication technology, that is, mirroring system replication data is a database transaction, belongs to the application layer, so as to ensure the integrity of the database data; The implementation does not need to stop the business system, suitable for 7X24 hours of continuous operation of the business system; ● No need to guarantee the consistency of the main and standby system hardware, greatly reduce the cost of system transformation and input of hardware, just a little more backup storage space on the line; The consumption of network bandwidth is very small, no special optical fiber transmission network, the use of real-time incremental replication technology greatly reduces the cost of resources, the impact on the performance of the service system is very small; ● Once the primary system fails, because the database of the standby system is in the running state directly, there is no need for data recovery phase, only the service system can be restored, so the whole process of the standby system to replace the primary system is very fast; ● Real-time incremental replication technology is adopted to copy data to the standby system. When the primary system fails, the number of lost standby system is very small. Due to the small amount of data, the backup window tends to zero, which has little impact on the performance of the primary system; ● Application mode diversification, support many to one, one to many mirror mode, to provide an extension platform for the later; ● Web-based unified management platform, responsible for the configuration of the server, database, set mirroring policy, and monitor the replication chain running, convenient management. You can monitor backup jobs and running status of the entire network on the WEB monitoring page. Each monitoring record is a complete replication link, including mirror agent, mirror server, and running status.

2. The advantages of the database Dr System based on the application-level CDP technology are as follows: ● ACDP understands the meaning of each data block in the application system, so that the application system can be restored to the consistent state and always ensure that the recovered application system is reliable and available. This is the characteristic of application-level CDP; ● Real-time data replication, continuous, uninterrupted any changes in the source end of the data real-time replication to the target standby end, not only to ensure real-time data synchronization, but also to ensure zero data loss; ● Real-time recovery: the client recovery software recovers the latest application data to the standby server. When the primary server fails, the standby system can be started immediately to realize the standby system without recovery. ● A-CDP stores data blocks one by one so that data can be manually restored to any point in time to achieve fault tolerance. ● Set recovery plan, automatically recover application system, can build a query system separated from the production phase; ● THE AUTOMATIC and intelligent synthetic incremental version of CDP storage makes the recovery path of full plus incremental plus small data blocks available, reducing the recovery path by 99%, thus greatly speeding up the recovery speed; ● CDP storage reserves the full or incremental version of history for recovery, which can replace the traditional backup function. You can set the number and duration of reserving historical versions based on the available storage space. ● ACDP uses ACA engine, during the implementation of the business system does not stop, does not affect the business system; ● The application mode is very flexible, can achieve a multi – to – one disaster recovery, one – to – one disaster recovery, to provide an extended platform for the later.

Two, multiple protection scheme

The scheme features are as follows: 1) What you see is what you get. The service databases on the standby end and the primary database are the same and visible in real time. 2) Online application-level DISASTER recovery. The standby end is online in real time and can be used without recovery. 3) Real-time data replication: production data can be copied to the standby database in real time without data loss; 4) The standby end can be queried, the standby end database is in running state, and the query and statistics service can be realized; 5) Implement no downtime, suitable for 7X24 hours continuous operation of the business system, does not affect the business operation. 6) The online Dr On the secondary end does not need to be recovered and can directly take over from the primary database. Other Dr Methods require a lengthy database recovery, and the database may be damaged and cannot be started, so the Dr Function is lost. 7) If the secondary end is online, users can use the service system to connect to the Dr Database at any time to verify the Dr Result. 8) For online Dr On the secondary end, the Dr Database can also be used for query and statistics to realize service distribution.