Has anyone tried setting up two physical servers running OpenSolaris 2009.06 HA Cluster to present NFS (or possibly iSCSCI targets) to ESX 4.0 hosts? I'm particularly interested in whether the failover works transparently as far as ESX goes and whether performance has been adequate on the VMware side. If so, do you have any configuration resources to which you could point me?
My OpenSolaris 2009.06 server is configured with the following:
Dell PowerEdge 1950 w/1 quad-core Xeon, 64 GB of RAM, Dell PERC 5/e RAID card, Broadcom embedded dual GB NIC
Dell PowerVault MD1000 w/2 x Intel X-25E SSDs, 13 x Seagate Barracuda ES2 750GB SATA drives
zpool uses 11 of the 750GB drives in RAIDZ2, 2 spares, and slog to the SSDs (PERC 5/e has each drive setup as RAID0 since it doesn't do JBOD naturally).
Performance appears to be pretty good as a standalone NFS Server for my ESX 4.0 update 1 servers but I need redundancy and am comparing this against HP Lefthand SAN/iQ on three IBM x3650 Servers. I'm considering replacing the 750 GB drives with ones that are 2 TB. That would solve my storage problems for a long time.
Thanks in advance,