The total time required to copy data (steps 2 to 7 in Performing data copy) can be estimated by the following formula. The processing time depends on the number and capacity of the Google Cloud Hyperdisks to be migrated. If an Google Cloud Hyperdisk includes unused areas, the data copy processing time is shortened because unused areas are not written to the migration destination volume.
5 [min] + (Number of applicable virtual volumes × 1 [min]) + Processing time for copy I/O [s] (1) / 60
If data is migrated from one Google Cloud Hyperdisk, you can calculate the processing time for copy I/O (1) by using the following formula:
Capacity of the Google Cloud Hyperdisk to be migrated [MiB] / (Read throughput of the virtual volume [MiB/s] + Write throughput of the migration destination volume [MiB/s]) (2)
The approximate estimated time (2) is 25.6 [MiB/s].
The preceding approximate estimated time assumes that Duplication Mirroring configuration is used and the data reduction function of the migration destination volume is enabled. The estimated time might increase or decrease depending on the configuration in use or whether the data reduction function is enabled. Also, the following conditions might affect the estimated time.
-
I/O load
-
IOPS and setting value of throughput for the migration source Google Cloud Hyperdisk
-
IOPS and setting value of throughput for the drive
-
Number of drives on a storage node
-
Number of storage nodes
-
How the free capacity differs among storage controllers
-
When the fault domain to which an initiator node belongs differs from the migration-destination fault domain
You can calculate the actual time by using performance information of the virtual volume and migration source volume after starting data copy (step 2 in Performing data copy) and before resuming operations on the migration destination server (step 6 in Performing data copy).
For details about how to check performance information about individual volumes, see Obtaining performance information about individual volumes in the VSP One SDS Block and SDS Cloud System Administration.
If data is migrated from two or more Google Cloud Hyperdisks, you can calculate the estimated time by using in the preceding formula the values for the Google Cloud Hyperdisk that has the largest capacity among those Google Cloud Hyperdisks. (Data migration can be parallelly performed for no more than 21 Google Cloud Hyperdisks.) If data is migrated from 22 or more Google Cloud Hyperdisks, data copy for 21 Google Cloud Hyperdisks starts first. When data copy for a Google Cloud Hyperdisk finishes, data copy for one of the remaining Google Cloud Hyperdisks starts. In the same way, data copy for the remaining Google Cloud Hyperdisks is performed one by one.