in reply to Re^4: Parallel::Forkmanager and large hash, running out of memory
in thread Parallel::Forkmanager and large hash, running out of memory
The relevant doc is here.
The tl;dr: version is: any number of processes can read at the same time. When a process wants to write to the database, the engine is notified, waits for current readers to finish, then gives exclusive access to the writer process. No one can read while a write is occurring. When the write is finished, everybody can start reading again.