stanonik@NPRDC.NAVY.MIL (Ron Stanonik) (02/03/90)
We recently needed to connect from a milnet site (nprdc.navy.mil) to a cerfnet site (sdsu.edu) , and found delays were about ~30secs. This is particularly annoying given that the sites are less than 5 miles apart. The milnet/arpanet core gateways were redirecting nprdc.navy.mil through reston-dcec-mb.ddn.mil; ie, both sites are in San Diego CA, but at least one site was being redirected through VA(?). Admittedly the milnet is only 56kb and is now an unfashionable backwater of the networking galaxy, but something seems very wrong with the routing to produce delays of 1/2 minute. Aside from the milnet noc, is there anyone else to query about routing problems (an nsfnet noc?)? Gone are the days when we only needed to sort out routing with one or two gateways. Is there an online description of the nsfnet topology? Not that it will help. Thanks, Ron Stanonik stanonik@nprdc.navy.mil
prue@VENERA.ISI.EDU (02/06/90)
>We recently needed to connect from a milnet site (nprdc.navy.mil) >to a cerfnet site (sdsu.edu) , and found delays were about ~30secs. >This is particularly annoying given that the sites are less than >5 miles apart. The milnet/arpanet core gateways were redirecting >nprdc.navy.mil through reston-dcec-mb.ddn.mil; ie, both sites are >in San Diego CA, but at least one site was being redirected through >VA(?). >Admittedly the milnet is only 56kb and is now an unfashionable >backwater of the networking galaxy, but something seems very >wrong with the routing to produce delays of 1/2 minute. I coordinate the routing within Los Nettos in the Los Angeles Area. Los Nettos and CERFnet have well coordinated routing and are well interconnected. They can be thought, topologically, to be the same regional network. They are not administratively the same but again are well coordinated. Los Nettos advertises sdsu.edu to the Arpanet and in particular to the mail bridge MARINA-DEL-REY-MB.DDN.MIL ==> 26.6.0.103 10.6.0.22 This mail bridge is co-located with a Los Nettos node and on the same PSN, #22, on the Arpanet. Within Los Nettos/CERFnet the route back is via full T1 to this mail bridge. That is, the return path does not touch the NSFNET backbone. I have no tools to tell me the path from your node towards sdsu.edu. Right now the ping times from the Los Nettos node co-located with the mail bridge, to sdsu.edu is averaging 182 ms. I do know that MARINA-DEL-REY-MB.DDN.MIL has been unresponsive to pings at various times and may or may not have forwarded packets. Walt Prue 213-822-1511
stanonik@NPRDC.NAVY.MIL (Ron Stanonik) (02/06/90)
Thanks for the info. We're now getting redirects from the milnet/arpanet core to route through marina-del-rey, instead of reston. Coincidently(?) rtt's are down to 500msec, which makes the connection usable. Hmm, maybe someone unthreaded a routing knot along the way? Thanks, Ron stanonik@nprdc.navy.mil ps. If anyone's interested, this is the traceroute from nprdc.navy.mil to sdsu.edu: traceroute to sdsu.edu (130.191.229.14), 30 hops max, 40 byte packets 1 MARINA-DEL-REY-MB.DDN.MIL (26.6.0.103) 241 ms 125 ms 152 ms 2 10.0.0.22 (10.0.0.22) 180 ms 185 ms 170 ms 3 ucla-gw.ln.net (130.152.64.2) 184 ms 394 ms 273 ms 4 CERFnet-GW.UCLA.EDU (128.97.10.253) 169 ms 226 ms 172 ms 5 sdsc-ucla.cerf.net (134.24.101.100) 402 ms 266 ms 261 ms 6 134.24.220.101 (134.24.220.101) 477 ms 520 ms 630 ms 7 * * * 8 * * * 9 * * * 10 * * * 11 * * * 12 * * * 13 * * * 14 * * * 15 * * * 16 * * * 17 sdsu.edu (130.191.229.14) 451 ms ! 615 ms ! 815 ms !