I'm basing this on the 2950 and 2820 routers that I use, so it may be subtly different on your 2910.
If you log in to the web interface, select "Diagnostics" on the LHS, and then look at NAT Sessions Table, Data Flow Monitor and the traffic graph you should be able to get some more info.
On the traffic graph, look to see if your session count plateaus (levels off) at some large-ish number. This would suggest that additional sessions are requested but cannot be satisfied.
If you don't, but you see large plateaus on either of the WAN bandwidth graphs, that might indicate that the timeouts are due to your WAN connection being saturated - how fast is it?
The data flow monitor tool gives you a handy summary of the active devices, and the number of sessions each has active. You might find that lots of sessions are being consumed by a server (if you have a public server behind your router), or by some rogue machine. If so, you could set it as a True-IP DMZ host (menu: NAT/DMZ Host -> Active True IP) if it is a server, or investigate the source of the connections if it is a client PC.
Another diagnostic would be to compare the success of ping from your laptop to the success of the ping tool in the router's diagnostics. If the router one works substantially better than your laptop, that would support the issue being down to overloading the router.