For now, when we use multiDB feature, the DBs are queried sequentially wich lead to poor performance when the number of DB arises.
It should be usefull to make the queries in parallel on each DB of the pool.
By example, an algorithm like "K-WAY merge" (https://en.wikipedia.org/wiki/K-way_merge_algorithm) could be used to accelerate the retrieving of results.
Require a significant level of effort so not currently planned.