RE: PC Routers (was Re: /24s run amuck)
Alexei Roudnev wrote: Purchase SuperMicro U1 server, with 2 9 Gb SCSI disks (hot swappable).
Suddenly that cheap router ain't cheap anymore.
Now, say, announce A crash Cisco IOS. 99.9% Internet backbones are Ciscos, so this announce breaks few Ciscos around and die - so you never know about it (and will not have a chance to be happy that _this announce crash Ciscos but do not crash ZEBRA). Not bad, of course - you are alive, all Internet is alive. Now, say, announce B crash ZEBRA (and do not crash Cisco). It will spread until it reach first ZEBRA on it;'s road - _your_ ZEBRA. So all Zerbras in Internet crash at once (and you are unhappy).
Another variant: announcement C crashes vendor "A", but not vendor "B" and not Zebras either (put whoever you want for "A" and "B" but there's only two of them on the backbone, mostly). Unfortunately, it takes a few minutes to crash, so it has enough time to propagate all over the Internet before the first "A"s begin to crash. As more "A"s crash "B"s will quickly be overwhelmed and the entire internet soon is down because no matter who coded it, when half of the backbones takes a hike the other half follows. Your Zebra is still up, but it does not do you any good because the entire Internet is down and you're only a small leaf on its edge, so nobody knows that you are still up, and they don't care anyway because no matter what they can't get anywhere. And yes this happens; not to beat or pick on any of the parties, I remember the entire AT&T frame relay network being down nationwide for more than 24 hours (for parts) because (as it is rumored) someone pushed a bad piece of software on a Stratacom switch. [me puts the asbestos suit on] <flame_bait> <FUD> If you have vendor C or vendor J, and all vendor C or J routers crap out at the same time, you're safe. Yes, you were down but so was half of the rest of the world, so it's obviously not your fault but vendor C or J's fault. If you have a zebra one a homebrew PC and all Zebras crap out: 1. You can't blame C or J. Worse, you can't blame anybody else. 2. As a coincidence, a sales droid from C or J will see your boss in the following days, take him/her to a nice restaurant, and will inevitably say "this would not have happened if you had a real router instead of this garage-built crud". 3. The sales droid is full of it, but _you_ are deep into it. </FUD> </flame_bait> Life is not fair, is it? Michel.
--- Michel Py <michel@arneill-py.sacramento.ca.us> wrote:
If you have vendor C or vendor J, and all vendor C or J routers crap out at the same time, you're safe. Yes, you were down but so was half of the rest of the world, so it's obviously not your fault but vendor C or J's fault.
Michel.
But this doesn't reflect the way the problems tend to spread: I've seen cases where something which crushes C gets injected, carried by Js across a network, and trashes all of the Cs in the network. However, it didn't spread to other providers, because the problem was { too many /32s | weird masks | an IGP messup | a J bug } For a problem to spread to other networks, it has to be perpendicular to the actual BGP configs, because most carriers apply just enough filtering on their peers to keep garbage like that out. Problems like that seem to be mostly customer-initiated. The ones that spread seem to be M$ related... -David Barak -Fully RFC 1925 Compliant- ===== David Barak -fully RFC 1925 compliant- __________________________________ Do you Yahoo!? Yahoo! Hotjobs: Enter the "Signing Bonus" Sweepstakes http://hotjobs.sweepstakes.yahoo.com/signingbonus
participants (2)
-
David Barak
-
Michel Py