Advanced Format Repository Disks & Instant Restore Overheads

Discussions specific to VMware vSphere hypervisor

Advanced Format Repository Disks & Instant Restore Overheads

Veeam Logoby Stoo » Mon May 09, 2016 2:08 pm

Trying to get a quick sanity-check on my understanding here from you clever folks:

VMware still doesn't natively support bigger Advanced Format/512e disks, but the extended capacity is obviously mega-useful in backup repositories.

If I make use of AF disks in a repository, though my hardware RAID controllers will abstract a lot of the physical drive structure, Windows 2012r2 still sees the underlying physical size as 4k ("fsutil fsinfo ntfsinfo %repositrydrive%" confirms Bytes Per Sector 512, Bytes Per Physical Sector 4096)

Question: How likely is this to significantly impact instant VM restore performances? I'm guessing that as W2012r2 (my repository/NFS host) is AF-aware, I won't run into any of the offset alignment issues that can cripple performance, but I would still potentially be impacted by the read-modify-write overhead if I modify a 512b block within a 4k sector? Where does this stand me in terms of VMware support - would an instant-restore NFS repository backed by AF drives technically be unsupported by them?

I've generally steered well clear of using anything AF/512e anywhere within an actual VMware production environment because of the lack of support from VMware's end so haven't really deviated from 'best practice' - anyone have experience of benchmarking AF stuff in a lab to see how much the performance knock is on the read-modify-write overhead? I realise to an extent it's linked to the IO profile of the workload, just keen for any anecdotes :)

Service Provider
Posts: 5
Liked: never
Joined: Fri Aug 23, 2013 8:42 am
Full Name: Stu P.

Return to VMware vSphere

Who is online

Users browsing this forum: Google [Bot] and 1 guest