Open phaus opened 9 years ago
It wouldn’t be too hard. But why would we want to do that?
If we want to use this software as an example for scaling over instances, having a Database in the Backend and a LB in the Frontend would be nice. Not that important though
Very good point, at the moment, scaling it by firing up additional instances would kill it :-)
We can essentially pick any datastore we like. Given that, I suggest we go with one that scales horizontally. A simple key/value store would be sufficient, a doc db would work as well.
The datastore itself needs to be refactored using a protocol (Clojure’s interface idiom), one implementation would use the existing file storage, the other one the DB.
Any suggestions? Riak? Mongo?
I would like to do the protocol refactoring as this is on my "things to do in statuses" list. But if anyone else has time feel free to do it ;-)
Cassandra would be a good fit, wouldn’t it?
The last time I played with it, Cassandra was a nightmare to set up. Has this changed?
docker run …
:-)
at least that won't be a problem for that specific use-case, since we can drop Cassandra into Mesos as well: http://mesosphere.github.io/cassandra-mesos I just need a proper application configuration.
Those who do the work should be the ones making the decisions, so go ahead.
Since I do have only limited Knowledge about Clojure, how difficult would it be to use a real DB (like MySQL) as a Backend?