Hi,
We have been advised that for a small cluster (8 nodes, ~50-60 VMs) it is perfectly acceptable to install VMM and SQL on a single highly available VM on that same cluster without any issues. I've been through the Microsoft best practice guides, adhoc documentation and a lot of forum posts and cannot find anything official to support this position, apart from lab environments.
Does anyone have a definitive answer, or can at least point me to documentation that confirms Microsoft will support this in the event of 'issues'. I've been told it's common way to deploy VMware management infrastructure, but is it simply a case of people assuming that SCVMM will be 'fine' if deployed in the same way? Are there not situations where you can cause something of a chicken and egg situation where you could misconfigure the network virtualisation on the cluster (for example) and by doing so mess up the VMM connectivity at the same time - the same VMM server you'd need working to fix the problem you've just caused?.....
Cheers, Rich.