The mirroring among servers is suitable for the operation if the amount of data and the amount of data update is small for the following reasons:
Why the mirroring among servers is suitable for the operation if the amount of data is small
When a disk is replaced, the data of the disk is copied to the replaced disk so that the disk is mirrored.
The operation can be continued during copying, however, availability of the operation becomes low. When the data amount is large, copying processing takes time and it leads to the longer period of the low operational availability.
For the required time for copying processing, see "3.9.6 Required Time for Copying Processing."
Why the mirroring among servers is suitable for the operation if the amount of data update is small
In the mirroring among servers, disk data is written via the network. In this case, the write performance is lowered compared to the environment without mirroring.
When an I/O error occurs due to the network error, the data difference occurs among the target disks of the mirroring among servers. If the network is restored, the resynchronization copying is automatically performed, and the data in the area where the data difference occurs is copied. If the amount of data update is small, the equivalence can be restored for a short time.
Note
The resynchronization copying which is executed after the following events copies the entire slice.
When an I/O error occurs in the memory area of difference information inside the private slice
When the resynchronization copying in process is canceled due to shutting the node, switching the cluster application, the sdxcopy command, etc.
If the amount of data and the amount of data update of the operation is large, use the shared disk.
