SRDF/ Sync and Metro - LUN sizes for VMware, how do you deal with that?

Dear community members,

 

I am a storage guy looking for fresh ideas regarding VMware environments and PowerMax LUN sizes when SRDF/S and SRDF/Metro are involved.

 

Old concerns/rules about SRDF LUN design I used to follow:

Stripped metas were preferred over the concatenated ones, to avoid IO queueing because of IO concurrency.

Many LUNs would perform better than a huge one with the equivalent size.

 

Challenges to deal with:

Huge virtual machines are taking place and the Vmware guys do not like extents inside datastores, they prefer huge LUNs instead of  many with the equivalent size, which is the totally the opposite to that I mentioned earlier as  “old school rules”.

 

We have a VMAX 250F with 2 vBicks, replicating against another VMAX 250F through 4 x 1Gbps DWDM channels, which are preforming well (low latency and BW below 20%)

Up to now, we have defined an arbitrary MAX LUN size of 4TB for the SRDF replicated ones without any IO contention issues. Actually we never tried going ahead with greater sizes.

 

How do you usually deal with this kind of situation? The LUN size is a concern for you? Which sizes are you handling today and how does it go?

 

Thanks in advance for you time reading my post, it is really appreciated.