I recently sat our 'VMware vSphere on NetApp (VVNA)' course for my instructor certification on the course. We use ESX 4.1 with an 8.0.1 7-mode set of simulators for the labs. It all worked well.
We get a lot of feedback from students when teaching our courses. I've had a number of students running ESX 4.1 telling me that they've been upgrading to 8.0 to take advantage of 64-bit aggregates. The biggest drawback would seem to be the implied performance hit with 64bit aggrs, but I haven't heard of any bad experiences with performance. I've even had a few students who had upgraded to 8.01 to take advantage of the larger ASIS volume sizes. 'After you upgrade to Data ONTAP 8.0.1 and later from an earlier release of Data ONTAP, you can increase the deduplicated volume size up to 16 TB. ' In general folks seem to be running deduped volumes on 32 bit aggrs for their VMs, and using 64 bit aggrs for data used by those VMs ( large DBs ). Not much more detail I could give you, it's just classroom discussion feedback.
Do realize that 8.01 is still in RC status when thinking about upgrading, and read through the release notes.
We are running ESXi 4.1 and OnTap 8.0.1 7-mode on a fas3270 and IBM HS22V blade servers with 10GBe passthrough using NFS to connect. We are also running ESXi 4.1 and OnTap 7.3.3 on a fas3170. We are seeing extremely high write latency with OnTap 8.0.1 7-mode. Read latency is fine. Has anyone else had this issue? Are there any known compatibilty issues?
What will P4 do for me? Is there something in P4 that will specifically address this issue?
Everything is up to date and configured according to Best Practices. We have identical config with our FAS3170 running 7.3.3 with no performance issues like this. We also have a N6040 that was running 7.3.2 and we recently upgraded to 8.0.1 7-mode and it is reporting the same write latency issue and was not before the upgrade.
We have been running 8.0.1 for quite sometime, only 64bit aggregates, on vSphere 4.1u1, and have not had any write latency issues. What are you guys seeing in terms of ms delay on writes to the array? do you have Operations Manager data readily available?