In general the notion of pool exists to collect pre-allocated resources which are expensive to re-create.

In Nuxeo we use a “db” pool for JDBC connections, which are expensive to recreate because they trigger network access and take time to execute in the SQL server.

And we use a “vcs” pool for low-level Nuxeo sessions, which are expensive to recreate because they hold a cache of already-fetched objects which is itself expensive to re-create.

And yes each time you use a Nuxeo session you have to take a session from the “vcs” pool, and also take a connection from the “db” pool.