We’re going to rely on disk snapshots for backups and also ship features for quickly cloning disks (or restoring from backup).
Our goal is to build plumbing to make things like Postgres/MongoDB/other HA work and then make the actual app projects open source.
The Postgres HA cluster app is what we’re running for this, which means two things:
- You can make pull requests to enhance functionality
- You can fork it and run it without going through our launcher
One (unique) thing we’re doing is giving out replica privileges on Postgres clusters. You can, for example, create a Postgres replica that streams to S3. We’d love to roll all that into our primary Postgres HA project, but you shouldn’t have to wait on us to do it. 
All our persistent volumes are nvme drives, each between 2 and 7TB. You can expect to get a guaranteed proportional amount of IOPs (500GB is about 25% of a 2TB NVMe capacity) with much higher bursts. You will probably be surprised how fast these disks are.