Unsolved
This post is more than 5 years old
2 Posts
0
1095
September 19th, 2017 10:00
eNAS on Vmax3 - thick or thin Tdevs?
Hi all,
We've been having an internal discussion about LUN / Tdev allocation for eNAS on Vmax3 arrays. I come from a Celerra/VNX background where it was invaraibly a case of 'provision thick LUNs for your NAS mapped pools' - especially if using thin filesystems which we do.
So, as eNAS is basically the same engine but virtualised, I've assumed the same principle applies with Vmax 3, albeit you have to 'fake' Thick LUNs by setting the 'preallocate' and 'persist' flags when you create Tdevs
However one of our TAs says he's been told by an EMC techie (can't remember who) that on Vmax3 that you only allocate your Tdevs thin so has insisted that I investigate further. Raising an SR has been, well, inconclusive on the question - IMHO I think terminology is getting in the way, as all Vmax3 Tdevs are nominally 'thin' whether you preallocate or not.
Can anyone give me a definitive answer on whether or not we should preallocate / persist Tdevs for eNAS mapped pools please? There is no 'eNAS best practice' document out there as far as i can see, just about all the NAS manuals are VNX-specific and/or getting on in years and the only explicit reference I can find for Vmax Tdev creation for eNAS is in 'eNAS File Auto Recovery with SRDF/S' (page 39) which says 'allocate thick'
Also - for AVM what's the recommended minimum LUN increment when you grow a mapped pool? The manual says it will stripe across 8, 4, 2 and 1 LUN increments but that doesn't mean doing so is a good idea :-) I know that the ideal is to always use, I think, 8 but other factors may not allow me that luxury :-(
Thanks,
Martin