Unsolved
This post is more than 5 years old
2 Posts
0
30675
August 28th, 2015 13:00
VRTX, M630, Hyper-V config questions
We are a fairly small shop, running on hardware that is up to 12 years old (with OS' to match!). We made a decision (prob the wrong one, but I'm trying to move on from that) to buy a VRTX with 2 hosts and around 12 Tb of disk in the DAS. We are planning on replacing 11 physical servers with this config, and virtualize most of them into this environment. The servers are configured to take the load of what we are replacing, so I'm not really worried about that, just the config part (no, we don't have any money to pay for config services either) :( :(
Was initially going to set up as a HA Hyper-V cluster, but have since realized that's just too complex, and not really what we need. Everything is in the same rack, connected to the same switch, using 2 different UPSes, just didn't make sense for us. Part of the 'moving on speech earlier'.
hardware config is as follows:
- 2 of dual proc M630 Blades with 16Gb of RAM each
- 1 of R2401 1Gb Switch module (ports gi0/1 and gi0/3 are connected to Corp switch, gi0/2 & gi0/4 are not connected to anything, should they be?)
- 10 Tb usable disk installed in the DAS (currently 2 partitions, one for each server)
Questions I have at the moment (new to Hyper-V, and been looooonngg time since I did any switch configuration).
- As this is 'just' going to be DAS-connected-to-servers, do I need to configure VLANs for this? (we are not going to have automagic VM failover capability)
- If I connect gi0/2 & gi0/4 to Corp network also, will that give the blades increased throughput? or will it just cause us to have to configure VLANs? (which is fine, I'm trying to do the best with what I have, just need to know)
- Each blade sees the 4 internal Ports on the 2401, yay! When I'm configuring Hyper-V, I am under the impression I need to configure 1 virtual switch for 3 of the ports and assign the 4th to the Hyper-V private network, is that correct?
- Or do I need to configure it differently? I was thinking I'd have to use teaming, to aggregate the connection, but the external is still only 1Gb, so it won't make a difference, right? can you tell Hyper-V networking is a challenge for me? I've read everything I can on it, and still makes me crazy trying to figure it out.
If I've missed anything obvious, (besides the platform booboo) please also let me know about that too.
thanks so much for your time!


Daniel My
12 Elder
•
6.2K Posts
0
August 28th, 2015 17:00
Hello
Are you referring to the Shared PERC as the DAS or do you have a NAS/DAS connected to the VRTX? If you are referring to the drives within the VRTX connected to the SPERC then they are routed directly to the nodes. They do not go across the network switch. All of their communication is contained within the VRTX.
Maybe. You can connect all of the ports to your network and should not have to segregate them with VLANs. Each physical port should have it's own IP and MAC address, so there should not be a spanning tree issue with the ports being disabled in any way if you connect all of them.
You will have increased network bandwidth available, but how that is utilized will depend on your configuration and the method of communication.
There is not a need to configure it any specific way. VLANs are for traffic segregation. If you don't need to segregate traffic then you could have one physical NIC assigned to the HOST OS and the rest assigned to a virtual switch to be used by your VMs.
If you don't have the budget for a consultant or the need for any advanced configuration then I would not bother with VLANs. Just leave everything on access mode and set to the default VLAN setting of 1 or VLAN disabled.
Thanks
_isemec
2 Posts
0
September 1st, 2015 13:00
thanks so much Daniel for your reply!!! I really appreciate it.
SPERC:
sorry, shared PERC is what we are using. I'm calling it a DAS, cuz we are not leveraging the capabilities of the card, just using it to access disks installed in the VRTX. Without configuring clustering on the Windows side, is my current configuration correct? We will get corruption etc, if I was to configure the entire array of disks as a single entity and configure partitions for the 2 servers, right?
Network:
That's what I was hoping, if I plugged in gi0/2 & 0/4, and assign them addresses, the installed servers will have access to them via the internal switch. we have no need to segregate traffic, at least not at this point.
thanks again for your response,
Daniel My
12 Elder
•
6.2K Posts
0
September 2nd, 2015 10:00
No, if you are not going to be clustering then you can't do that. The nodes need their own independent storage. If you create a single storage pool and just try to partition it to the various nodes it likely won't work at all, but it will cause corruption. You can have a shared storage for shared data, but the OS and applications need to be installed on a drive that is just assigned to the node unless you are clustering.
Thanks