2 small or 1 big server?
Hi Alex and Doug,
> We have solid reconnect code built into our software. Seriously–if you’re working with a network connection expect it to degrade and fail.
> We solved this problem by killing connections on the backend when they’ve been idle after 10 minutes. This means a client who logs in, works for 30 minutes, then comes back 4 hours later isn’t using a backed connection for most of that time.
> We?ve kept our customer databases separate (like alex) and have reconnection code (like alex) that handles most situations except for death of the connection in the middle of a transaction.
Thank you for your advises. Leaning towards 1 big server…
But we do not have an API yet, so how do you handle reconnection then? Our attempt to do this results in hanging for a period while negotiating for connection.
Our simple and dirty solution is to activate keepalive to avoid reconnection delays, but this means also a bunch of idle connections.
I have read that pooling skips the ”negotiation” part of the connection. Just picking a free session.
Any reconnection code that you are willing to share?
Manage your list subscriptions at lists.omnis-dev.com