Migration Tool

As mentioned above, MatrixDB uses the mxshift tool to perform data migration operations.
The mxshift tool currently supports the following features:

  • Full migration: implement parallel migration of data from Greenplum 5 / Greenplum 6 / MatrixDB cluster to a new MatrixDB cluster in database units
  • Reverse migration: Reverse migration from MatrixDB to Greenplum 5 / Greenplum 6
  • Incremental migration: Incremental data migration through WHERE conditions
  • Replicated Table

mxshift logic schematic: ![](https://img.ymatrix.cn/ymatrix_home/logical diagram_2.0_1679387458.png)

mxshift physics schematic: ![](https://img.ymatrix.cn/ymatrix_home/Physical Map_2.0_1679387464.png)

If the data migration process is likened by water flowing through the pipeline and the data transmission efficiency is likened by water flow rate, then the water flow rate is often limited to a low value as a whole because of the "narrowest part of the pipeline", resulting in the so-called "single-point bottleneck". With this in mind, the mxshift tool implements data transfer from the Segment directly to the Segment, eliminating single point bottlenecks that may occur in common migration operations. When the topology of the source cluster and the target cluster are inconsistent, that is, heterogeneous, mxshift can also calculate the optimal mapping scheme based on the number of source cluster segments, the number of target cluster segments, whether the segment is on the same physical machine, etc., to realize automated data migration.

Notes!
For detailed parameters, please refer to mxshift