VMware with NFS

Hi Guys,

I'm putting together a solution for a school with 2 sites and what they're terming a DR cabinet. This is a juicy one I'm afraid as I have a few questions.

Being a school they have a large number of users with reasonably low user concurrency and a limited budget. I'm positioning the VNXe to meet a price point however iSCSI is dealt with via the datamovers iSCSI replication and snapshotting incurrs a large overhead (essentially 100% of the LUN + the TWS should these snaps be mounted and written to). My understanding is that NFS only requires that the Savvol is sized to accomodate the rate of change much like snapview and does not require this 100% provision upfront. Obviously the number of checkpoints and duration these checkpoints will remain for also comes into play. Also, my understanding is that replicator V2, the Snapsure Savvol is used for replication checkpoints also. does this also mean we can save some capacity there ?

Essentially I'm just looking for some clarification that this is the case and I'm not going to find that I've largely undersized the storage to accomodate the Savvol.

Also with NFS3 being quite chatty and constantly checking to see if file systems are still there and if they still have certain files etc, with a barrage of GETTATTR requests. In real terms what are the implications of using NFS in a VMware production estate (all be it for a smallish server count of 15-20VM's). I understand that there are specirfic timeout settings to be tuned for NFS and that by setting NFS in uncached mode for VMware we can mitigate this, but does anybody have some end user/real world experience of this ?

One final question with respect to SRM to manage the underlying replication. I understand that SRM for now supports many to one replication and there are 2 sides to this question. Firstly at the target how is this achieved, do we require a Vcenter instance for each source ? also, what are the considerations from a Celerra replicator perspective ? can we implement NFS with Replication manager for machine consitency and still use SRM 4 ?

Apologies for the barrage, I am digging through white papers as we speak and not just throwing questions out expecting answers.