Effect of changing timeouts to 3 hours

Your persistence tables will grow, which will require more resources.

Also, consider whether it's beneficial to extend persistence timeouts beyond the session timeout on on the servers.

Thanks Arie. I'm wondering if there is any formula or way to calculate just what that growth may be

The growth would depend on the number of clients connecting, and what kind of persistence is used. It's not just a matter of number of clients, as many clients could share the criterion for persistence (e.g. IP-address).

I would start by looking at the size of your current persistence table. However, before that you may want to determine whether it even makes sense to extend it to such a long time, especially if the persistence would exceed the session timeout on the servers. In that case I'd question whether anything can be gained.

Avoid using excess timeout values if possible. Often longer values are requested because users complain of being on a system idle for a long time and the LTM drops their connections because they were idle. If this is the case then it is better to have the client send some sort of keep-alive regularly so that the entries do not expire.

Two examples I would point to is SSH can have keep-alives enabled and HTTP can have a very small java script embedded that does a request for a very small object every so often.

Changing the timeout values significantly will mean you are always using those resources where fixing the problem with some sort of keep alive fixes only for those that have long idle sessions and keeps your resources allocated to current clients.

Different methods of persistence will have different uses of server resources and different user impact.