Hyper-V Clustering

S

Soldier101

Gawd
Joined
Jan 8, 2002
Messages
639
How do you guys configured your Hyper-V clusters?

I have three physical servers and 2 sans.

I want the three physical servers in a cluster so that if one fails the Vm's move from one to the other two.

I keep reading on Hyper-V Clustering for this but it sounds like it is guest clustering vs physical and I'm getting confused.

Would like all devices to hit the same SAN and use it for the same storage etc.
 
we will be connecting to the store using ISCSI. I basically have no idea how to configure this off the bat lol
 
I am wanting basically to have one giant store that the servers access via ISCSI. That store will hold the VM files. But I keep reading things about Quorum (spelling?). and making sure they are a cluster to prevent two machines from trying to write to the same block etc.
 
Yes, look into iscsi failover clustering.

You need a small quorum disk that acts as a tie breaker/arbitrator. You also need to make sure your SAN supports simultaneous IQN connections.

The idea is that you connect all of your hosts to the same volumes in the iscsi snapin, then let the failover cluster actually mount the volumes.
 
Thanks. I will see if I can figure that out. I have some documentation I am looking at. The storage can support it. They are the ones that referenced me to look at clustering in the first place so that works out.
 
Hyper - V Clustering requires the physical boxes to be joined to a domain? That's what I'm reading and that is what has me super confused.
 
DO I need to build a quorum on the SAN itself or do I let Windows create that? I'm so damn lost lol
 
Soldier101 said:
Hyper - V Clustering requires the physical boxes to be joined to a domain? That's what I'm reading and that is what has me super confused.
Click to expand...

Yes you need to be in an AD domain.

Soldier101 said:
DO I need to build a quorum on the SAN itself or do I let Windows create that? I'm so damn lost lol
Click to expand...

You create a volume for the quorum disk. That disk acts as the vote tie breaker in the even of a failure.
 
Only thing I really have left is the SAN failover. I have 2 SAN's. I want to make it to where they back each other up (if one fails the VM's switch to the other). I know I need them to replicate and stuff. It is the Hyper V configuration I am struggling with on that part lol....VMWare was soo much more intuitive
 
Soldier101 said:
Only thing I really have left is the SAN failover. I have 2 SAN's. I want to make it to where they back each other up (if one fails the VM's switch to the other). I know I need them to replicate and stuff. It is the Hyper V configuration I am struggling with on that part lol....VMWare was soo much more intuitive
Click to expand...

Well that is an operation aside from your cluster. Depends on the SAN. I run Dell Equallogics and they manage all of that without the cluster ever knowing.
 
What SANs do you have?
Are they the same make/model & do they have the same drive layout?
Typically if you're going for SAN-SAN failover you're paying premium to the SAN vendor. There are 3 ways to configure the replication.
1 - configure the replication feature provided by your SAN vendor. in this case you'll get a replicated LU (iSCSI/FC volume) on both SAN nodes. It will be visible to all 3 initiator cluster nodes as 1 LU and MPIO will figure out the rest. Whenever SAN 1 fails, MPIO switches path to SAN 2. The same with broken paths to the SANs. Connect both SANs to your Hyper-V host as per your SAN vendor's manuals and then create the failover cluster. Schizrade already mentioned the domain so that should already be taken care of.

2 - if your SANs are dumb and can't replicate. You can use StarWind Virtual SAN software on your Hyper-V nodes to replicate between the SANs and also boost performance. This option is also true HA like the first one, meaning that replication is synchronous and failover is automatic. This is something I'm doing to 2 older VNX boxes for one of my customers so I can go into a little more details.
The idea is to give entire SAN1 capacity to Hyper-V1 and SAN2 to Hyper-V2. Then use StarWind on the Hyper-V nodes to grab the SAN volumes and mirror them. StarWind calls it HA Device and it's configured using any volume available in the system. When you create the HA make sure you give it at least 1-2 GB of Write back cache, your VMs will thank you for that. RAM cache in their case is replicated between the nodes so you can't lose your data if a node fails.
Once configured you get 2 StarWind targets, one on each Hyper-V node. You want to have at least 2: 1GB quorum/witness and whatever the size you need for your VMs. You then go to iSCSI initiator on all 3 Hyper-V nodes and connect all Hyper-V hosts to the StarWind targets. They do have a manual here: StarWind Virtual SAN: Installation and Configuration of Hyper-Converged 2 Nodes with Hyper-V Cluster. This one covers your scenario except for the 3rd Hyper-V node, but it's going to be easy as you just connect the StarWind targets there and you're good. The guide actually covers the cluster part so I won't duplicate it here.

3. Hyper-V replica. In this case you configure each SAN as a separate volume on your Hyper-V nodes. SAN1 is the primary storage, SAN2 - failover. Keep in mind, this option is one-way and is not really full auto compared to the 2 options above. You'll get a Hyper-V cluster running all VMs on SAN1, then you configure Hyper-V replica for the necessary VMs and point the wizard to the CSV residing on SAN2.

I'll be glad to get into more details so let me know if you have any additional questions.
 
You must log in or register to reply here.
Back
Top