RAID and Windows licensing

s3. · Tue Mar 06, 2018 9:25 am

Hello,

I am planning on testing StarWind VSA, I have a few questions though:

1. As I want a big all-flash setup (6TB+ each server), I was wondering if there are any consequences not using local raid on each server? As you will understand, such big SSD SAS/NVMe disks are really expensive. I already searched the forum regarding this and I know RAID-1 over network is possible, but I'm not sure if there are any other caveats.
2. How would I need to license the nodes (2) regarding Windows Server 2016? If I have 38 cores in each server, would I need to license them both or only one?
3. Regarding hardware to choose: if I have a given VM environment which would require a given amount of resources of RAM/vCPU/IOPS, should I split those resources over 2 nodes or do I need to have the same amount of resources on 2 nodes? I know that if you split the resources and 1 node would fail, it would stress the node which is live, but my question is only regarding the scenario when both nodes are live.

Thanks in advance.

s3.

Tue Mar 06, 2018 2:23 pm

Hello s3,
Let me try to answer your questions.

1. As I want a big all-flash setup (6TB+ each server), I was wondering if there are any consequences not using local raid on each server? As you will understand, such big SSD SAS/NVMe disks are really expensive. I already searched the forum regarding this and I know RAID-1 over network is possible, but I'm not sure if there are any other caveats.

You can find recommended settings for the storage there. In case of all-flash, you can set RAID 5 for the arrays.

2. How would I need to license the nodes (2) regarding Windows Server 2016? If I have 38 cores in each server, would I need to license them both or only one?

You would need two Windows licences, for both servers. The type of the licences depends on your needs and the amount of VMs you will run.

3. Regarding hardware to choose: if I have a given VM environment which would require a given amount of resources of RAM/vCPU/IOPS, should I split those resources over 2 nodes or do I need to have the same amount of resources on 2 nodes? I know that if you split the resources and 1 node would fail, it would stress the node which is live, but my question is only regarding the scenario when both nodes are live.

It is better to have two identical servers because in this case you can distribute your VMs and their load between two servers. In case of one of the failure one of the nodes your production will proceed to work from the second node.