From icg09146@noc.icg.net Sun Feb 13 09:33:38 2000 Date: Tue, 8 Feb 2000 09:40:20 -0800 (PST) From: Morgan Collins To: noc-staff@noc.icg.net Subject: Re: help (fwd) Ralph, thanks for the note. Now I realize the error of my ways. ******************************************************************************* ---------------------------------------------------------------------------- Morgan Collins, Network Administrator icg09146@noc.netcom.net ICG Communications, Inc. ---------- Forwarded message ---------- Date: Sun, 6 Feb 100 23:33:06 -0800 (PST) From: Ralph Pearson To: Morgan Collins Subject: Re: help I would say it would have to be a combination of factors. The latency to Dallas is too high (should be in the 30-40 range), but the increment from Dallas to Miami is too high as well. Miami to Tampa looks ok, though. Check out the health of the trunkage between L.A. and Dallas, and between Dallas and Miami, and I think you'll find the culprit(s) somewhere. Is routing for said trunkage effecient? Is utilization at a comfortable level? Are the circuits supporting the trunks error free (i.e. is layer-2 ok?)? How is the CPU load on the devices supporting those two links in the chain? Lastly, don't count on traceroute (even matt's) to give you an accurate picture of what payload traffic is subject to; Cisco routers prioritize icmp (and in particular, the icmp types assocaited with traceroute) as lowest priority -- so you may see absolutely abysmal latency numbers in a traceroute that follows a path that is virtually unaffected by any adverse conditions other than relatively high CPU load. While this is inconvenient for us in terms of troubleshooting, it makes some sense: if you are traceroute'ing, you are trying to determine what path your packets take, not necessarily whether they are getting there expediently; similarly, typical ICMP traffic is geared towards verifying connectivity in general, not guaging quality of service, so as far as a Cisco is concerned, as long as the packets get through and a response is offered, the Cisco's cycles should be spent processing payload as a priority before they do anything more with ICMP. > From icg09146@noc.icg.net Sun Feb 6 23:20:31 2000 > Received: by noc.noc.icg.net (8.8.5-r-beta/8.8.5/(ICG v1.2)) id XAA29903; Sun, 6 Feb 2000 23:20:29 -0800 (PST) > Received: from localhost (icg09146@localhost) > by orca.noc.icg.net (8.9.1/8.9.1/(NETCOM v2.00)) with SMTP id XAA05882 > for ; Sun, 6 Feb 2000 23:20:29 -0800 (PST) > Date: Sun, 6 Feb 2000 23:20:28 -0800 (PST) > From: Morgan Collins > To: noc-staff@noc.icg.net > Subject: help > Message-ID: > MIME-Version: 1.0 > Content-Type: TEXT/PLAIN; charset=US-ASCII > > Hostname %Loss Rcv Snt Best Avg Worst > 1. 209.111.91.150 0% 92929 92929 0 0 > 16 > 2. spinaltap.icgcomm.com 1% 92919 92929 0 0 > 53 > 3. sjr-ca-gw1.icg.net 1% 92921 92929 1 1 > 581 > 4. s11-1-0.sji-ca-gw1.icg.net 1% 92920 92929 2 13 > 1959 > 5. a10-0-0-1.lai-ca-gw1.icg.net 1% 92918 92930 10 25 > 1777 > 6. a3-0-0-2.dai-tx-gw1.icg.net 1% 92912 92930 64 107 > 2222 > 7. a5-1-0-4.mig-fl-gw1.icg.net 1% 92299 92929 98 147 > 1601 > 8. tam-fl-gw1.icg.net 1% 92450 92929 107 155 > 1612 > > > I was looking at tampa florida, and I can't really seem to find where this > latency is coming from. > > ALSO: HOW THE HELL DO I LOG OUT OF THIS COMPUTER, IM A LITTLE LOST. > > --m > > ******************************************************************************* > ---------------------------------------------------------------------------- > Morgan Collins, Network Administrator > icg09146@noc.netcom.net > ICG Communications, Inc. > > > -- Ralph Pearson Director, Network Operations Center ICG Communications, Inc. (San Jose, CA) 408-579-5222 / ralph@noc.icg.net ........................................!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!