I know this is going to sound crazy, but finally going to virtualize my old server to new server we have. We are mostly cloud based apps, so that is the reason for just now moving. I have used Hyper V at my old job over 5 years ago, but I have a question on best practice for new setup.
I have new Dell server, and I bought with thought of having 2 drives raid 1 for main OS and Hyper V Manager, and having the virtual machines on separate Raid 5 on another set of disks. Does this sound like best course of action?
I’ll second Chris’ question. But also wanted to add: Make sure you’re using a good hardware RAID controller. If you’re using a “light” or software RAID system, you will NOT get good performance on that RAID 5 and it will be very obvious within your VMs.
I also think RAID5 should be used only with serious consideration in virtual environments. You’ll be hitting those disks harder than a normal NAS or file server, so they’ll be more likely to experience failures and RAID5 doesn’t give very good likelihood of recovery when dealing with modern large capacity rotational drives. In general, I’d rather go down a grade in drive (Say, from 10k RPM to 7.2K RPM SAS) and do RAID 10 if I can meet my capacity needs in the budget for the improved resiliency and overall performance.
I was wanting way to have better disaster recovery and get server back up quicker to another box. Was thinking run host OS and Hyper V on 2 Raid 1 disks, and setup the virtual machines across the Raid 5 disks (one file server and children’s checkin virtual, and one for anti-virus console). All hard drives are SSD on Dell hardware Raid, and have had issues in past where ran one standard file server and did Raid 5 across drives, and OS corrupt and couldn’t boot and almost lost data drive that was on same Raid 5. So started running separate container on diff. drives so OS was completely independent of data Raid container
I tend to use KVM personally given I don’t have to run as much Windows Server dependent stuff, but a good idea with disaster recovery, especially for the VMWare and Hyper-V world is to look at a VM backup product like Veeam as you can do a disaster recovery, or even a fail-over if it’s a really critical load, out to the cloud and if things really go sideways you can always keep running on Azure or AWS until on-site is resolved. There are a few other vendors out there too but I haven’t worked with them much and I’ve never met anyone who didn’t like Veeam (I’m sure they are out there though).
If you are using hardware based RAID vs. software based your configuration should be fine. In fact running that exact configuration with 1 host and 12 virtual servers and it’s running just fine.
If these are critical systems that can not loose information, possibly consider RAID 6 vs. RAID 5. Same concept as raid 5, but you can loose up to 2 drives before any data is lost.
As long as you maintain good backups, treat RAID as a notification system for when a hard drive fails. What I mean by this is if you trust (and test) your backup system, it is easier on your drives if when a drive fails, you reformat the raid and restore the VMs from backup. This will save unnecessary read/writes on your drives. (Rebuilding a RAID array on a failed drive is hard on existing drives, especially if they are SSD drives with a finite number of read/writes.)