From your URL: ** NICs are standard 1 Gbps **
That's a bad idea :( Invest into RDMA capable 10 GbE hardware and it would definitely pay back!
Technology Tips and News
From your URL: ** NICs are standard 1 Gbps **
That's a bad idea :( Invest into RDMA capable 10 GbE hardware and it would definitely pay back!
If you want anything serious, you would invest into 10Gbit cards, as Anton suggested.
My guess is that you can 'squeeze' MGMT network into 100Mbit, but to have 900Mbit for Production and heartbeat together is just not enough I am afraid.
Maybe at start (in case you have 2-4 VMs, not utilized much) it'll be enough.
And for Live Migration, is it only one card, no redundancy?
I would invest into 10Gbit cards, reserve 1Gbit for MGMT, 7Gbit for Production, and 2Gbit for Live Migration.
These are just my 2 cents.
I won't be quite as harsh as stating that you 'need' 10 GE, but there is value. That said, I have run labs and small environments that do not require performance on a single NIC. Do things work? Yes. Is it optimal? No.
There is no specific 'heartbeat' network in clusters. It is known as the 'cluster communication' network. That carries the heartbeat and other cluster communication traffic. It's not a good idea to run the cluster communication traffic on the SMB network unless that is your only fallback. It will work, but it is not recommended. Since you have a teamed NIC for production/management, that will automatically be selected for cluster traffic. The teaming will ensure you have a secondary path. You don't really get to select a primary and secondary for cluster communication. If a NIC is selected for its use, it can be used.
Without knowledge of loads, that's really about all that can be said. You will need to monitor to see if your NIC speeds are sufficient for your needs.
Hello,
Nice opportunity for a good laugh here..potentially.
Looking to deploy an SOFS and HV cluster over the coming summer. Going through the network configuration and have come up with the following (link to image below). Each cluster is comprised of 2 nodes, so nothing too heavy on the network side of things.
SOFS cluster will utilise SMB 3.0 app shares, on a JBOD.
Would welcome suggestions/bad mistakes/judgement calls..
Thanks for all of the replies and suggestions.
Can anyone suggest any techniques/software for producing a compiled baseline of performance, for our existing virtualisation setup?
Many thanks.
"Can anyone suggest any techniques/software for producing a compiled baseline of performance, for our existing virtualisation setup?"
Microsoft Assessment and Planning Toolkit - www.microsoft.com/map
"Can anyone suggest any techniques/software for producing a compiled baseline of performance, for our existing virtualisation setup?"
Microsoft Assessment and Planning Toolkit - www.microsoft.com/m