Прочети на българскиIn a storage environment with OpenZFS the easiest mistake is to think only in terabytes. The real question is which layout supports the required workload, what the recovery expectations are and how the system behaves under simultaneous read/write operations.
In our publicly described architecture the storage layer is separate from the compute layer, and that is a deliberate decision. It makes it clearer where the risk lives and where discipline is needed when making changes.
What really matters
- latency under a real workload, not just synthetic numbers
- snapshot and backup logic that is understandable and maintainable
- a clear distinction between performance storage and backup storage
- a plan for how you recover when something breaks at an inconvenient moment
We will also publish a separate topic about backup and replication logic, because it should not automatically be mixed with the storage performance discussion.
Related topics
Want to see these settings in action?
All our VPS services are built around the same engineering approach. If you are looking for real infrastructure rather than just a reseller service, see the VPSStorages plans and infrastructure.
For more background on the technology see what ZFS storage is and why it matters.