we are usign piwik in near 30 websites and we are getting around 100gb of raw data each 15 days.
It is not very efficient to have input data, deletion jobs and aggregation jobs pointing to the same database. Deletion and aggregation demands a lot of I/O which is causing that we are reaching the limit of concurrent sessions (which is very high) at peak hours. Also is very normal to find unfinished aggregations at the portal, for example we have hourly data from some projects but not daily data.
The point is, we already have a MySQL replica in order to send some custom/advanced queries to de raw data. Would be a killer functionality (in terms of performance) to be able to configure the source of the data to be used for aggregations and probably the summary data for the portal from the replica.
With this functionality we would have more healthly databases and also different paramaters for each database in order to improve different db work.
by the way, is feasible at this moment a manner to hardcode that?