This topic provides information that you should take into consideration for FlashCopy mappings.
Ensure that you've considered the type of I/O and frequency of update before creating the virtual disks that you wish to use in FlashCopy mappings.
FlashCopy will perform in direct proportion to the performance of the source and target disks. That is, if you have a fast source disk and slow target disk, the performance of the source disk will be reduced as it has to wait for the write to happen at the target before it can write to the source.
The FlashCopy implementation provided by the SAN Volume Controllercopies at least 256K every time a write is made to the source. This means that any write will at a minimum involve a read of 256K from the source, write of the same 256K at the target, and then a write of the original change at the target. Therefore, when an application is performing small 4K writes, this is translated into 256K.
Due to this overhead, consider the type of I/O your application will be performing during a FlashCopy. Ensure that you will not be overloading the storage. The calculations contain a heavy weighting when FlashCopy is active. The weighting itself depends on the type of I/O being performed. Random writes have a much higher overhead than sequential writes, for example, as the sequential write would have copied the entire 256K anyway.
You may spread the FlashCopy source virtual disks and the FlashCopy destination virtual disks between as many managed disk groups as possible. This will limit the potential bottle-necking of a single storage subsystem, (assuming that the managed disk groups contain managed disks from different storage subsystems). However, this may still result in potential bottle-necking if you wish to maintain all your target virtual disks on a single storage subsystem. Ensure that you add the appropriate weighting to your calculations.