 |
Several conveniently selected disks instead of one single disk can speed up the sequential read/write performance to
more than a triple. Use the logical volume manager to set up the disks.
Avoid using subsequent disk addresses in a storage server (e.g. the addresses 5100, 5101, 5102,..
in an IBM TotalStorage Enterprise Storage Server 2105), because they use the same rank.
- This makes it slow
- All disks are from one rank and accessed via the same path.
-
- This makes it fast
-
- Use as many CHPIDs as possible
- Spread the host adapters used across all host adapter bays
- If both an FCP CHPID and the host adapters (HAs) support the same link speed, connecting one CHPID with one host adapter results in the best performance.
Otherwise take care that the link speeds of interconnected CHPIDs and host adapters match.
- Spread the used disks equally over all ranks and both clusters
- If possible, avoid reusing the same resources (CHPID, host adapter, rank)
This applies to FCP and FICON.
|
Assume, the following items are available on a ESS 2105:
Then assign these resources in the following way:
For FICON: Define a path from each CHPID to each LCU.
|
|
|
|
|
