webber@porthos.rutgers.edu (Bob Webber) (06/18/88)
In article <10382@ulysses.homer.nj.att.com>, smb@ulysses.homer.nj.att.com (Steven Bellovin) writes: > In article <5017@dasys1.UUCP>, tneff@dasys1.UUCP writes: > > 1) If AT&T changes from a high volume backbone site to a high volume > > leaf site (this is how I interpret their announcement anyway), then > > what lucky non-AT&T machine gets to be their feed? ;-) Were the > > administrators of said lucky machine consulted during all these > > "high level meetings"? ;-) > > This should actually cut the traffic. Look at it this way: no traffic > that flows in to the gateway will flow back out -- it will be seen as > duplicate, and bounced. 1) Bounced means sent back out, just out the same way it comes in -- so bouncing in and of itself won't be a particular improvement. 2) This move might have made sense prior to ``smart'' mailers, but the net has been abused by ``smart'' mailers now for over a year. This means that the only mail heading thru ihnp4 was that that was locally generated (at a minor site that hasn't yet bothered with a smart mailer) or mail that a smart mailer had determined should be going thru ihnp4 because that was the cheapest route. With the conversion of the AT&T network to be reaching the outer world only thru att.com, no one would know of any of the internal links so the only time att.com would see pass thru mail would be when it was the shortest path between two sites that it directly connects to. Now those two sites are going to have to talk to each other directly (which was obviously more expensive for them than talking to ihnp4 or they wouldn't have used the ihnp4 connection in the first place). Net effect -- non-att mail will now be more expensive for the net as a whole and most likely for sites that neighbor att as well. ------ BOB (webber@athos.rutgers.edu ; rutgers!athos.rutgers.edu!webber)
steve@edm.UUCP (Stephen Samuel) (06/22/88)
From article <Jun.18.03.15.21.1988.6317@porthos.rutgers.edu>, by webber@porthos.rutgers.edu (Bob Webber): > In article <10382@ulysses.homer.nj.att.com>, smb@ulysses.homer.nj.att.com (Steven Bellovin) writes: >> In article <5017@dasys1.UUCP>, tneff@dasys1.UUCP writes: >> > 1) If AT&T changes from a high volume backbone site to a high volume >> > leaf site (this is how I interpret their announcement anyway), then >> > what lucky non-AT&T machine gets to be their feed? ;-) Were the ... > Now those two sites are going to > have to talk to each other directly (which was obviously more expensive > for them than talking to ihnp4 or they wouldn't have used the ihnp4 connection > in the first place). Net effect -- non-att mail will now be more expensive > for the net as a whole and most likely for sites that neighbor att as well. It's not necessarily cheaper to go thru att than direct -- at least, not in phone charges. However, it often IS cheaper to do it in terms of administration. (why get a direct connection when ihnp4 has all the connectivity??). There was a time when the fastest way to get a message from Edmonton to Calgary (about 180mi) was to bounce it thru Michigan and Vancouver (twice across the continent)!! Needless to say: a direct call to Calgary would have been cheapest, but the way things were set up, it wasn't the fastest. -- ------------- Stephen Samuel Disclaimer: You betcha! {ihnp4,ubc-vision,mnetor,vax135}!alberta!edm!steve BITNET: USERZXCV@UOFAMTS