If you needed any convincing about the benefits of VAAI, here’s a graph of what happened when I took our new VAAI-capable HDS AMS 2500, copied a 25 GB template VM to it, then cloned three more VMs from that template. I did the cloning one at a time, rather than in parallel, mainly because I was in shock that it took about 30 seconds for each one to complete and in my giddiness I didn’t think to do any other testing. I have only the Hitachi write rate highlighted, but that tells the story:
So far, the only drawback I can see to the new VAAI offloads are just that you’ll need to rely more heavily on your array’s performance monitoring. Which may also mean that you’ll have to work with your storage admins. I know, I know, that’s scary and difficult, but, to borrow the words of Arista Networks’ Doug Gourlay at the Gestalt IT Networking Field Day, you guys are just going to have to hug it out.[0]
——————
[0] He was speaking of networking guys vs. virtualization guys, but it’s the same thing. This “virtualization guy vs. everybody else” attitude has got to go away.[1]
[1] People who know me personally know I’ve been very critical of storage admins in the past, and I’m not being hypocritical here, as there’s more work I have to do on this front as well. But it’s getting better, and will be the subject of another post.
Comments on this entry are closed.
That’s where I’m lucky – I’m the virtualisation guy *and* the storage guy. :-)
Sweet performance… Will use it to convince my customers ;-)
You’re Cloud Admin (Admin). No more Storage or Virtualisation Guys.
That is cool :-)
Cool stuff isn’t it. We also have on the VSP and USP-V(M) now so complete portfolio covered. :-)
Sorry, couldn’t resist.