Hyper-V Migration/LUN Sizing

[DrCheese]

Hi all,

 

I'm moving from VMWare to Hyper-V over the summer this year - Migrating the VM's to a standalone Hyper-V machine, reformatting the main SAN, replacing two of the servers in the cluster & then transfering the VM's back to the main SAN/Cluster.

 

At present our SAN is set up with 4 LUNs of 3TB each - Giving us around 12TB. This has caused us some hassles in the past with servers struggling to fit in that space (i.e our main data server is over 3TB so it has drives split across two LUNS.

 

What's the best practise for this? i.e Is there a reason why I can't just have one big LUN or two 6TB ones? (I'm aware I'll need a additional smaller LUN for a witness disk) The only reason I can think of is that we once had a corrupted LUN on vmware that we had to clear out and reformat....

[Steve21]

Was talking to some of the Dell storage guys about this when we replaced our SAN and basically as you said no there isn't a limit "but" they advised us to only use the minimum we need 2TB LUNs for us, as if one gets corrupted/breaks it won't take your whole network out (There's also the comment on if any of your servers are old OS's whether it'll support it but assuming this would be 2012r2?)

 

Steve

[maddocks1993]

Hi Dr Cheese,

 

I believe there is no best practice in order creating numerous Volumes on a SAN, reason being is because it is down to yourself or whoever will be setting the volumes up. It is mainly to do with your backup, sometimes if they Volume has been configured as one LUN then the backup may take longer, it may not. Same goes with Hyper-V, as long as you can find Storage for Hyper-V whether it's iSCSI / SMB / Fibre Channel it will be fine. You could NIC team the nodes and put an extra switch in LCAP or use the new SMB 3.0 Application Share for Hyper-V (Microsoft Recommends Now) and I have seen more people start to use SMB 3.0.

 

Dell told us to create numerous volumes on our Tier-2 incase you wanted to replicate Volumes to another SAN. All SANs work the exact same way, so it all depends how you want to configure your SANs. As long as you have got Hot / Cold Spare disks available in the correct Raid you are fine my friend as one disk were to go down the hot spare should carry on the process whilst that broken disk gets replaced. Dell are a very 'Middle Man' I see them as, they just sell products but they rarely give you advice as they always say have you looked on Microsoft or this website. Again, they will go down the route of IOPs as well. They did that our place and told us we need this amount of IOPS when we only used a 1000 IOPS which a Raid 1 NAS could do quite easily. Keep options open as well, HP are the best for SANs / Servers. Hyper-Convergence, Nutanix (Dell & Lenovo uses Nutanix Software) , HP , Gridstore are mainly good for Hyper-V.

 

I will give you a run down on what we have setup with our Hyper-V Cluster.

 

3x Nodes running Hyper-V

2x EqaulLogic via iSCSI to the 3x Nodes. 1x on a Raid10 and 1x on a Raid6

Failover Cluster installed and configured to use Virtual Machines for High Availability.

Cluster Volumes are Tier-2 & Tier-3. Meaning all disks in SAN are just one Volume per EqualLogic with Correct Raid Level and Spare disks.

NIC Team using LCAP or use SMB 3.0 Application with Live Migration set as SMB.

 

We are using Veeam to backup our Virtual Machines which are done on the Hyper-V Level (Block Level). If you aren't with Veeam I would highly recommend them because the things you get with Veeam Backup & Replication is just amazing and you could restore your entire Network Infrastructure backup in a matter of minutes depending size of the Virtual Hard Disks.

 

Chris.

Edited June 2, 2016 by maddocks1993