04.10.2019 15:50, Mark Rotteveel wrote:

> The problem is inherent to the monitoring tables. If you query one
> monitoring table, Firebird needs to collect and prepare a snapshot of
> the data for all other monitoring tables to be able to provide a stable
> and consistent view of the monitoring tables in your current
> transaction. If you do this a lot, then it will need to repeatedly spend
> that time, which can have an impact on performance, because the cost of
> that data collection is non-zero (.
>
> So, the problem can't really be fixed, because that data collection will
> need to happen and doing the collection takes time (CPU) and memory.

Just an addition:

In v2.5, pending MON$ requests could also [temporarily] block other
connections' operations, thus decreasing the overall performance. This
was improved in v3 to reduce the cross-locking with other connections.

But generally, your message is absolutely correct.


Dmitry