Hyper-V? Not in my datacenter… (continued)
In 2008 I wrote an article named “Hyper-V, not in my datacenter!” which was based on Hyper-V version 1. In 2009 I wrote an article named “Hyper-V, the laughter continues” and “ESX vs Hyper-V mythbusting myth” which was based on Hyper-V version 2.
Microsoft promised/warned us that with Windows Server 2012 Hyper-V version 3 this would all be history. But as Sander wrote in a more recent article named “Windows Server 2012 Hyper-V: overpromising and underdelivering” this turned out to be a marketing statement based on a premature product with no suitable management tools.
But now with the release of System Center 2012 Service Pack 1, Hyper -V version 3 should be the VMware vSphere-killer (according to MS).
Too bad, the sequel continues! Yesterday Edwin pointed me to a great article by Justin Paul, a fellow blogger, vExpert, VCP and EMC specialist, in which he describes his recent struggles with Windows Server 2012 Hyper-V.
True, in the last four point Justin refers to some information which is a bit outdated, but he has a valid point stating “Then explain why it happened to a 2012 hyperV cluster if it is fixed ?” when people in the comment section claim that Microsoft solved this in the new version of Hyper-V. Apparently they weren’t fixed.
When I read Justin’s comments on the CSV issues and the hassle to setup a Hyper-V environment, I stand by my earlier position, this is not a product which I trust to base my enterprise IT infrastructure on. So still, “Hyper-V? Not in my datacenter…”.
Posted this over to his article:
I’m not sure what happened in your situation and I totally don’t disagree with anything you’ve said regarding management and design, since it is Microsoft and shit happens. But I’ve been running our 2008 environment for a few years with little issues, 30 some hosts, 1000 some vms, 40 some CSVs.
Just a few things I’d like to point out:
1. “Cluster Shared Volumes have been around a long time, Microsoft has been using them for everything from Exchange clusters to SQL clusters.”
As far as I know and have setup, MSSQL has never used CSVs, just shared disks, since MSSQL works in an active/passive mode when clustered, the disk is only owned and put online on 1 node at a time.
2. In 2012 CSV’s are now supported in multiple roles besides Hyper-V, such as Scale-Out file servers. I’ve talked with a few of the people who are responsible for it and if I remember correctly, they stated you can put anything on it, but you know….
3. “http://www.ms4u.info/2011/05/why-you-should-not-running-domain.html <- Failover manager required AD, if you virtualize all your Domain Controllers Failover Manger wont start."
In 2012 you no longer need AD to be up to bring up CSVs.
4. "do a google search for HyperV live migration using multiple NICs"
Teaming should do that? We do it now. But the real problem is the limitation in 2008 with only 2 at a time, and general inherit slowness. It's much improved with 2012 from what I've seen personally.
5. "Look up how to do nic teaming in HyperV… Then look up how to do it in VMware"
Broadcom did have some issues 3-4 years ago, my last employer and current one experienced this. But using the vendor teaming tools has not produced any issues for me. I don't disagree that management in 2008 requires more steps. In 2012 it's rather simple, and can be done in SCVMM with SP1. I find SCVMM 2012's network fabric much easier to manage than vSphere personally.
That's probably not all, but I don't want to write a book either. I'm also rather new to a lot of this, so if there are any corrections feel free to smack me.