Sorry Alex, but I just don't get your point. > Think about it - with one or two 1G ports the usual flow is > still some hundreds Mbit/s, the typical load above 400 quite > often - too much for a public exchange already in my humble > opinion (per peer). Then one of the exchanges goes down, now > you send traffic to 5 peers with 150-250 Mbit/s each on > another exchange, together with some 2 other operators as > they have more than one peering with the same peers that you > have. Do you believe that everyone has their 1G ports at, > say, 100 Mbit/s only, just to have capacity free for the > rerouting of other operators? From experience, no, hardly. > > With private ports failover scenarios just work... you know > what you have, and you plan ahead. That is what I mean with > 'securing' traffic. If an interconnect breaks, I know who is > to blame. With some exchanges I feel secure, with some I do > not, and if .5 Gig traffic goes to other exchanges, how many > times did we overload the other peers' ports in the past? > > It all depends on your traffic levels. Yet was this hard to > work out? Rather not...
"it all depends on your traffic levels" - Well state the obvious Alex! You seem to forget that you +do not know+ what is behind the interface that you peer with whether its public or private. So one may have 4XSTM-4c interconnects with BIGPROVIDER only to find out that when one of them goes down BIGPROVIDER can't transit the traffic to one of the other ports. Its just another dimensioning issue on the network which is exactly the same issue whether you have a port private of public. The risk is _+exactly+_ the same. Regards, Neil. _______________________________________________ swinog mailing list [EMAIL PROTECTED] http://lists.init7.net/cgi-bin/mailman/listinfo/swinog
