YMatrix
Quick Start
Cluster Deployment
Data Model
Data Writing
Data Query
SQL Reference
Maintenance and Monitoring
Tool Guide
Troubleshooting
FAQ
We believe that in MatrixDB, data migration mainly consists of three phases: The first phase, environmental preparation; The second stage is migration execution; The third phase, follow-up tasks.
First, before the formal migration operation begins, you need to do a series of preparations, usually including but not limited to the following steps (adjusted according to the specific scenario):
Before stopping the business: back up old data, prepare the software environment of the target cluster, stop all DDL operations on the business side, back up the DDL statements required for migration, correct and optimize the DDL statements (recommended source information to create statements without indexes to improve migration efficiency) to adapt to the target database, etc.;
After the business is stopped, collect the software and hardware configuration information of the target cluster, compare and analyze the characteristics of the source cluster and the target cluster architecture to establish an appropriate migration plan).
We believe that these steps are necessary, and you need to ensure the security of old data and find the best way to migrate to reduce the cost and risk of stopping your business.
After the preparation work is complete, we enter the stage of performing migration operations. At this stage, we recommend not migrating the mirror node (Mirror), but adding Mirror after executing using the mxshift tool, which will greatly improve the migration efficiency. The mxshift tool will automatically print the migration progress according to the point in time when each table is migrated, and check the correctness of the migration in real time.
After the migration is completed, we recommend that you recreate the index and execute the analyzedb command in the entire library to update the statistical information of the new library, finally restore business access, observe business operation, and keep track of the data for a period of time (the specific time depends on the specific timing scenario), and finally complete data migration.
For specific steps, please see: Migrate data from Greenplum 6 to MatrixDB 4 Migrate data from MatrixDB 4 to MatrixDB 4