Gents,
I was going through the new document on MS Failover Cluster configuration on vSphere 5.1 and found one recommendation I could not understand.
it says "If you must overcommit memory, the swap file must be local, not on the SAN."
Can someone tell me what memory overcommitment affects swap file's location?
Thanks.
Its just a best practice to keep swap on local datastore, if you need to reduce using SAN storage space.
Hi,
You can find here a good white paper for explain Memory management on vSphere. : http://www.vmware.com/files/pdf/perf-vsphere-memory_management.pdf
Julien.
In many cases it's not so much an issue of space consumption on the SAN, but more about keeping the swap I/O localized, and more importantly, segregated from normal storage I/O. If you get to the point where VMs are swapping, it's often undesirable to have that I/O traversing the same fabric as your VM disk I/O, since it increases your risk of IOPS/throughput contention.
There are other reasons apart from MSCS to keep swap local, especially if you're using array-based replication over a WAN. Cost is another one, as Mustafa mentioned. I think in this case though, they're most concerned about keeping your regular storage performance as consistent as possible.