5

I`m using this S2D calculator, which takes reserve space into account. If I understood correctly, reserve space is only being used for in-place recovery in case a disk fails.

Is it correct that one is able to use this reserve space as storage at the price of missing in-place recovery?

I will use a 2-node cluster, so I think this should not be a problem, right?

Also, I currently have 6x1TB SSD and 2x2TB SSD. What is the best way to devide them between both nodes (for S2D only, no OS drive(s))

5

Is it correct that one is able to use this reserve space as storage at the price of missing in-place recovery?

Yes. You will be able to keep up and running while S2D rebalances itself using reserve space. Theoretically, this can save you from data loss in case you wasn't able to replace failed drive in time, and the second drive failure occurred after S2D finished rebalancing. Unfortunately, I didn't manage to find any detailed description of this feature and it's impact on the number of drive failures S2D can tolerate. The chances of loosing your data after the second drive failure are still pretty high.

Your disk layout will be as follows:

3x 1TB SSD per node: contribute into usable capacity. 1x 2TB SSD per node: will be used as reserve space.

You are looking at nearly 3TB usable capacity in such case.

P.S: I can't disagree with @BaronSamedi1958: 2-node Storage Spaces Direct is a weak setup. I don't recommend using it in any serious production.

There are alternative solutions that work better in 2-node configurations:

  • StarWind VSAN Free - works like a charm in 2-node environments, can be deployed on a bare-metal Windows.
  • HPE VSA Free - a good piece of software as well, but runs only inside a virtual machine, which leads to performance compromises. 1TB limitation in free version.
| improve this answer | |
3

I will use a 2-node cluster, so I think this should not be a problem, right?

I think you'll have an issue. During one node being down say for maintenance (patch Tuesday?) you don't have anybody to sync your data with, so disk failure will represent catastrophic results. Same thing is going to happen in case of a double disk failure: one disk died and S2D rebalances it using free space, and now second disk dies and you have a data loss.

I'd strongly recommend to bring in at least third or better third and fourth nodes (yes, expensive!) for a fully fault-tolerant S2D cluster. Remember: Microsoft started with 4-node requirement, and brought 3-node & 2-node much later after numerous requests.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.