does valkey support speedb?

[michaelrun]

We'd like to use speedb as backend storage of redis, if valkey supports speedb, we'd like to try.
Can anyone share any information about this?

[madolson]

Do you have a more specific definition of your use case? We don't support any tiering to disk today.

[reconbot]

I'm not OP but I've had a few large production use cases for tiering to disk, the main use case is large datasets (lots of hashes, and lots of sorted sets) where SSD performance is good enough for data that isn't "hot" but operations and development is significantly easier if you can treat it like one dataset so you don't have to deal with caching and invalidation especially in distributed systems. Additionally cost of ram compared to SSD is significant.

Last I checked Redis Labs, Snapchat and a few others have bought up anyone working in the redis SSD space, AWS and Redis Labs both have closed source solutions where they keep values on a SSD. Doing this in an open source project that will remain open source is a fantastic opportunity.

[durlabhjain]

Cost is definitely one of the factors. In one of our apps, we are using SSDB (https://github.com/ideawu/ssdb) with NVMe storage. The compressed SSDB store size is 1.7TB and performance is quite good at fraction of RAM only solution.