[Okta Webinar] Learn how to a build a cloud-first strategyRegister Now

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 248
  • Last Modified:

Looking for suggestions on NetApp/Hyper-V design

I'm fairly new at NetApp and have recently installed one at each of 2 sites with the help of a consultant for site #1. What I have is as follows:

Site #1

FAS2220 with an attached DS2246 storage shelf (iSCSI)
2 standalone Hyper-V hosts (2012 R2)
CIFS shares for site # 1 user files

Site #2

FAS2520 with and attached DS4246 storage shelf (iSCSI)
2 standalone Hyper-V hosts (2012 R2)
CIFS shares for site # 2 user files

The controllers are configured for HA at each site but not clustered between the sites. I will be using SnapProtect to backup CIFS shares and VMs at each site and replicating the backups to the other site for offsite storage. A consultant initially helped me set up site # 1 as follows:

All 12 disks from the FAS2220 are assigned to controller A and there is a LUN created on there, which is mapped as a drive on Hyper-V Host 1 to house the VMs and VHDs for machines running on that host.

12 of the 48 disks in the DS2246 are assigned to controller B and there is a LUN created on there, which is mapped as a drive on Hyper-V Host 2 to house the VMs and VHDs for machines running on that host.

The remaining 36 drives are unassigned at this time (I added them recently and haven’t decided how to assign them yet).

Does this seem like the best case setup for this environment? It seems to me that in this scenario, if I wanted to shut down a host for maintenance I would need to do a live storage migration and a live VM migration. Or am I just not getting it? Any input, ideas, or recommendations would be greatly appreciated. Thanks.
0
rsgdmn
Asked:
rsgdmn
  • 3
  • 2
1 Solution
 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
Usually you would split the disks between the controllers, and have an Active/Active setup. (in the event of controller failure, the other controller would take over the failed controller automatically!)

So the Hyper-V hosts would be accessing LUNs of certain sizes to controller A and controller B at the same time. Size of LUNs between 500-800GB, or sizes suitable for your VMs.

You are correct, if you need to shutdown a host for maintenance, you would Live Migrate the VMs from one host to another host.

You will need to create a Failover Cluster and HYper-V Cluster and Clustered Shared Volumes for this.
0
 
rsgdmnAuthor Commented:
Thanks Andrew.

I was expecting to have to do a live VM migration if I needed to shutdown a host as we chose against the failover cluster and CSV for many reasons, but I'm hoping I don't have to migrate storage prior to doing a VM migration.

So should the LUN for the VMs and VHDs be common between the 2 controllers and both of the Hyper-V hosts? I'm still confused on how Hyper-V Host 1 could do a live migration to Hyper-V Host 2 without doing a live storage migration if the VMs and VHDs are stored on a LUN only mapped to Hyper-V Host 1.

Maybe I'm just overcomplicating the thought process?
0
 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
Storage and VHDs can remain in place they do move, just the running process  moves!

All LUNs must be common to ALL Hosts!

thats what is usually done.
0
What does it mean to be "Always On"?

Is your cloud always on? With an Always On cloud you won't have to worry about downtime for maintenance or software application code updates, ensuring that your bottom line isn't affected.

 
andyalderSaggar makers bottom knockerCommented:
Of course you don't have to do a storage migration to shutdown a host, the storage on that site is still up and running you just move the VMs to the other host on that site. Someone said you were correct but actually meant you were incorrect.

Also "All LUNs must be common to ALL Hosts!" probably incorrect as well on a dual site setup, you certainly don't want the hosts on site A seeing the replicated LUNs on site B as well as their real-time data.
0
 
rsgdmnAuthor Commented:
OK, so each LUN created, although assigned to one controller, should be mapped as a drive with a common drive letter on each host? And then if I do a live migration of a VM it will access the LUN through the mapped drive on the host the VM is moved too? Or am I still confused on how this works?
0
 
Andrew Hancock (VMware vExpert / EE MVE^2)VMware and Virtualization ConsultantCommented:
You will need to use Microsoft Failover Clustering and create a Clustered Shared Volume, and then assign the same LUN/Volume the same driver letter.

You cannot connect two hosts directly to the same LUN, using the same drive letter concurrently, with disk corruption occurring.
0

Featured Post

Industry Leaders: We Want Your Opinion!

We value your feedback.

Take our survey and automatically be enter to win anyone of the following:
Yeti Cooler, Amazon eGift Card, and Movie eGift Card!

  • 3
  • 2
Tackle projects and never again get stuck behind a technical roadblock.
Join Now