At the moment I'm using ServerSession.deliver() for each message to remote clients, even in case of high volume of messages.
I'm wondering whether batching, as in:
- ... multiple deliver() calls
would perhaps work better.
The source messages (that I receive from multiple threads and that I'm routing through) could be arriving at random times, so I'd like to implement a sort of "time-out" mechanism: if I don't receive any message within x number of milliseconds, I'd like to send what's in the queue in one batch and start a new batch. This is intended for messages to a specific remote session only.
The goal here is to save processing overhead at the remote end, which we have seen to be a problem with a sufficiently large number of messages.
Is there perhaps a built-in mechanism in Comet to do this?