In my opinion, the term “SDS” is more loosely defined. I would say the basic function of an SDS platform is to manage storage resources (volumes) using software, independent of any specialized hardware.
Distributing data for a single volume across many nodes in a cluster, like a parallel filesystem or an SDS platform that supports erasure coding, is a feature of some distributed storage platforms out there. However, at this time, there are no plans for LINSTOR to support such features.
Besides the point that this is not how DRBD works today, aggregating storage in the way OP explained really wouldn’t be something anyone would want to do in production. If you wanted to get 30GB out of 3 nodes with 10GB each, then it’s enough for just one disk in any of the three nodes to fail for a complete loss of data. I don’t think that such a product even exists.
For the sake of comparison, if you used a platform that supports object distribution and space efficient erasure coding, you’d still have to give up about 33% of your space in a three node cluster. With a performance nowhere near to what you get with a fully replicated SDS. It’s for a different use case.
In a production environment, your explanation is correct.
However, portworx or the product in the following link provides this function.
These can be added to the total volume.
And replication between small volumes within the entire volume is also possible.
Therefore, I think it would be good for sale to be able to provide this volume summation function even if the customer does not use both functions at the same time.
A better deployment pattern for persistent layer would be to abstract storage devices available across multiple server nodes and create a virtual pool of storage, from which container schedulers can draw storage […]
In my understanding, that’s what Linstor does. The main difference I see is that replication is not an option but mandatory.
Yes. You are right.
The core of LinStor is replication using drbd. Replication is not an option; it is a core feature. LinStors provides simple and powerful storage HA features on k8s.
What I’m saying is, I thought that adding a volume aggregation feature similar to that of competitors would help LinStor’s sales activities.