Copy Data Management (CDM)
Data Disaster Recovery provides a sandbox solution for MySQL and Oracle based on copy data management (CDM) technology. You can quickly create a sandbox instance, such as an emergency RDS instance, to make your backup data readily available. Read and write operations in the sandbox do not affect the source database. The sandbox also supports recovery drills, development and testing, query analysis, and emergency disaster recovery.
Background information
In large-scale data scenarios, traditional database backup services face two growing challenges:
-
Storage cost: Periodic full and incremental backups are required to meet the recovery point objective (RPO). These backups generate large amounts of duplicate data, resulting in high storage costs.
-
Usage cost: Data can be used only after it is fully restored. For large datasets, restoration consumes significant network bandwidth, disk I/O, and CPU resources, and extends the recovery time objective (RTO) to hours or longer.
Data Disaster Recovery provides a complete sandbox solution for MySQL and Oracle based on CDM technology.
-
MySQL sandbox instance: Data Disaster Recovery provides storage and compute resources for MySQL sandbox instances and hosts their service processes. You can remotely access these database instances.
-
Oracle sandbox instance: Data Disaster Recovery provides storage resources for Oracle sandbox instances through a remote file system and hosts their service processes. The compute resources for these instances must be provided by your ECS instances.
The Data Disaster Recovery sandbox solution uses several advanced backup technologies. The following figure shows the technical architecture.
Benefits
-
Database physical backup: Backs up database files to significantly increase backup and recovery speed.
-
Data deduplication: Significantly reduces the storage cost of your backup data.
-
Storage snapshots and cloning: Makes backup data available quickly without copying data.
-
Storage virtualization: Provides sandbox instances that use cloud resources and are independent of your production instances. Read and write operations in the sandbox do not affect the source database.
Scenarios
-
Development and testing based on online data (Dev or Ops)
Testing and pre-release validation require existing online data, but this data must be protected from corruption. The sandbox quickly provides a copy of the full historical data to support rapid test iterations and ensure data reliability.
-
Online emergency disaster recovery
The sandbox-based backup system provides an RTO or RPO of minutes for instant recovery of online data. Compared with multi-active hot standby technology, the sandbox also provides data rollback capabilities to quickly roll back dirty data.
-
Disaster recovery drills
The rapid recovery capabilities and pay-as-you-go model of the sandbox support regular disaster recovery drills. This is ideal for scenarios that require high data reliability.
-
Ad hoc analysis of historical data
In many data analytics scenarios, you need to run ad hoc queries on historical data. The sandbox quickly provides a copy of the full historical data for these queries. Compared with existing batch analytics solutions on big data platforms, the sandbox provides the SQL interface of the original database, so developers do not need to write batch analytics scripts. The snapshot storage mechanism also keeps storage costs low.
Billing
For more information, see Sandbox billing.