On the peering of pTLAs...

[Bob Fink]

Alain's mail below on using RTT info for helping to determine optimal
backbone peering is a good one.

Has anyone thought of good ways/metrics to decide how pTLA sites can
determine what other pTLA sites to peer with to avoid the old backbone mode
of everyone just trying to peer with everyone?

My hope in all this conversion/backbone-cleanup is that we gain the
necessary reliability to have a reliable backbone.  pTLA sites need to be
ready to provide reliable service!  (sorry to preach :-)


Anyway, ideas and suggestions needed (to the list please), and appreciated.


Thanks,

Bob

=============================================
At 2:09 PM -0700 9/16/97, Alain Durand wrote:
>Networking is so fun !!! Especially in europe ;-)
>
>I'm trying to figure out who I should peer with in the new addressing plan.
>I know the connection from Renater/france to the US is &*^*&^*&^*&%^^%^&,
>so I'm looking mostly for european peers.
>
>It's late (11pm local time), networks should be quiet, time for some pings &
>traceroutes on the underlaying IPv4 topology.
>
>For each site, I'm giving the average RTT from G6, % of packet loss
>and the approximate route.
>
>IPv4 connectivity of 6bone core sites in Europe to G6:
>
>Site               RTT packet route
>                         loss
>----------------------------------------------------------------------------
>TELEBIT/DK         244ms   0% Ten34UK-stockholm-dk
>SICS/SE            50ms    0% Ten34UK-stockholm-se
>G6/FR                         LOCAL
>JOIN/DE            45ms    0% Ten34DE-frankfort-muenster
>SURFNET/NL         85ms   40% Ten34UK-stockholm-amsterdam
>IFB/GB             605ms  20% washington-sprint-insnet-ifb
>CSELT/IT           300ms  20% Ten34UK-stockholm-ParisEbone-interbusinessIT
>UUNET-UK/GB        200ms   0% washington-pipex-UK
>UNI-C/dk           80ms    0% Ten34UK-stockholm-dk
>ATT-LABS-EUROPE/CH 460ms   0% ten34Londre-stockholm-amsterdam-unisrc-att
>SWISS-TELECOM/CH              no valid record in database
>NETCOM-UK/GB       270ms  10% washington-netcom-netcomUK
>SWITCH/CH          40ms    0% ten34DE-ten34CH-switch
>JANET/GB           230ms  20% ten34UK-janet
>STUBA/SK           180ms   0%
>ten34DE-amsterdam-stockholm-munichEbone-bratislava
>INFN-CNAF/IT       50ms    0% ten34DE-ten34IT-infn
>INR/RU             2500ms 60% washington-sprint-dtag.de-moscow-inr
>
>If other sites had similar informations on their connectivity,
>I would be happy to get them to help me decide on the peering.
>
>	- Alain.