Scaling with Chrondb

[michal-kuzak]

How would this scale ? We did something similar year ago, and this proved over time a real problem. A lot of small files, without the possibility to scale or port. So what's the Architectural idea here ?

[avelino]

Scaling isn’t a concern yet, but based on the architecture, I believe writes will be slower since we need to persist data to disk

On the other hand, reads go through Lucene, so we shouldn’t have issues there, since the most accessed records are indexed (and will be in memory)