The first question which piece of equipment is actually failing.
Generally I would use 'tracert' for this, but in this case, the outage probably isn't long enough for this to be useful. Instead, we'll use tracert to determine the IP addresses of the possible points of failure, then use 'ping' to determine if these devices are up and running.
run 'tracert', and keep track of the outbound hops. Make note of the IP addresses of the first few lines:
$ traceroute google.com
traceroute to google.com (209.85.225.99), 30 hops max, 60 byte packets
1 router.localhost (192.168.1.1) 1.639 ms 2.627 ms 2.957 ms
2 10.0.0.2 (10.0.0.2) 13.848 ms 14.173 ms 15.637 ms
3 foodsl01.foo.02.gateway.foo.com (xx.xx.xx.xx) 23.273 ms 25.413 ms 27.746 ms
(Note that tracert is spelled 'traceroute' on my linux box... same principle applies)
Open 3 'cmd' windows, then start running a constant ping from each window to one of the IP addresses... in this case, I would start 1 ping to 192.168.1.1 (my linksys router), one to 10.0.0.2 (DSL Modem) and one to xx.xx.xx.xx (my ISP's gateway, which I've obscured).
Depending, on which of the IP addresses times out, gives you a place to start with trouble shooting... if your ISP is dropping the connection, you need to call tech support.
If it's the router
- check all cables reboot the router...
some routers become unstable over
time
- make sure that you have the most
recent version of the router's
firmware
- replace the router (I used to do tech
support on VoIP boxes, and I'm not a
big fan of Belkin)