Show HN: CurioDB – A distributed/persistent Redis clone built with Scala/Akka
github.comIt's good to see a solution that makes efficient use of CPU cores on each machine instead of forcing you to scale horizontally from the beginning ;p
I think a lot of servers/database engines today tend to focus on horizontal scalability across large numbers of single-CPU instances and tend to ignore the efficiency benefits of a hybrid approach using fewer large instances.
Hey Jon, long time no see. :-)
Yes I imagine the thinking there is similar in your project, eg: Redis/CurioDB :: Node/SocketCluster (https://github.com/SocketCluster/socketcluster).
We should catch up sometime! I'd definitely like try out CurioDB at some point.
That'd be good!
Cool project. Out of curiousity (and having no experience with scala or akka): is the lack of support for transactions and atomic Lua scripts a consequence of the overall architecture (one actor per K/V), or some other technical limitation, or just something you haven't gotten around to?
Yep, one actor per KV means some kind of distributed coordination would be needed - that doesn't exist yet.
Hey! This is interesting (&awesome), the other day I was looking at Riak and some Redis alternative written in Go, both had some different approach which wouldn't make the best use of persistencey and distrusted approach at the same time.
The performance table looks great, though what would be a recommended minimum underlying hardware requirement for production? (since the test is on a Mac-Air)
Glad you like it!
Hardware requirements would depend on your usage of course, it's like asking about the length of a piece of string. :-)
Honestly though, the code probably needs a lot more eyeballs on it before it's production ready, which is why I posted it here.
It looks good, am sure there will be lots of folks to help on the project.
Ah yes that is true regarding the usage/req :), but I thought you might have tried it in production with some requirement.