Setting: Small office with a mix of Ethernet printers, Ethernet Macs, and Airport Macs. About ten hosts.
Network config: Verizon DSL through a Westell WireSpeed modem, to an off-the-shelf Belkin router / switch / 54g access point, to a crufty old Asante hub, to a patchbay, to the walls.
Situation: Randomly, but quite frequently — usually once or twice a week, but it has even happened more than once a day — all systems will lose their connection to the Internet. LAN connectivity is unaffected, and I *THINK* I’ve seen certain Internet applications (retrieving mail from POP3, for example) continue to work, although I haven’t yet been able to verify that. HTTP definitely dies.
Remedy: In each case, the problem is fixed very easily by a soft reset of the router. I can even do this through the Web-based management, so obviously the hub is working properly ?- at least after the fact.
Evidence: Exhibit A: The hub is extremely flaky, but it doesn?t seem possible that it could cause Internet outages without causing LAN outages.
Exhibit B: The symptoms are time-outs, not failures to find the hosts, so that would seem to indicate that the DNS servers are still reachable?
Exhibit C: I kept a little log of several outages, and only once out of five was the WAN IP address leased by the router the same before and after the reset. In fact, there was a point where the address changed from 151.x.x.x, to 141.x.x.x, then stayed the same, then changed back to 151.x.x.x. I?m no expert on ISP-to-ISP routing or on DHCP configuration, but doesn?t that seem weird?
Finally, the question: Can I deduce from this what the weak link is? If not, what would be a good methodology to figure it out? Unfortunately, we?re a small non-profit, so I can?t throw money at the problem. (Otherwise I would have cut that hub out of the path some time ago.) So I want to at least rule out the possibility that this is all Verizon?s fault before I try a new router.
Thanks!