We're currently a NetApp 7-mode shop and have been testing Tintri, Pure and XIO. Today I noticed an interesting issue, sVmotion to NFS datastores on NetApp and Tintri get 300 to 350MB/s throughput, but NFS or iSCSI datastores to iSCSI on Pure and XtremIO only get 100MB/s to 150MB/s. I'm at a loss to explain why this would be the case. The iSCSI array all have MPIO configured, etc. and the vendors agree they are configured according to their best practice for all SATP/PSP/advanced settings. Confirmed no bandwidth issues on the switch ports across all systems, no network errors. Performance for regular VM activity has no issues.
Here are the run times for the same VM moved between different datastores.
FROM | TO | Start | Finish | Duration | Type |
XIO | Tintri | 12:48:19 | 12:52:16 | 0:03:57 | iscsi to nfs |
Tintri | Pure | 12:53:51 | 13:06:11 | 0:12:20 | nfs to iscsi |
Pure | Tintri | 13:07:57 | 13:11:52 | 0:03:55 | iscsi to nfs |
Tintri | XIO | 13:13:10 | 13:27:21 | 0:14:11 | nfs to iscsi |
XIO | NetApp | 14:58:04 | 15:02:04 | 0:04:00 | iscsi to nfs |
NetApp | Pure | 15:12:31 | 15:24:33 | 0:12:02 | nfs to iscsi |
Pure | XIO | 15:28:01 | 15:33:03 | 0:05:02 | iscsi to iscsi |
XIO | Pure | 16:12:02 | 16:21:18 | 0:09:16 | iscsi to iscsi |
Pure | NetApp | 16:26:55 | 16:31:28 | 0:04:33 | iscsi to nfs |
NetApp | NetApp | 16:34:23 | 16:42:08 | 0:07:45 | nfs to nfs |
NetApp | Pure | 16:53:49 | 17:07:10 | 0:13:21 | nfs to iscsi |
Pure | Pure | 17:12:56 | 17:16:29 | 0:03:33 | iscsi vaai |