Re: Peering versus Transit

[Bill Woodcock <woody () zocalo net>]

          Alan Hannan (alan () mindvision com) writes:
        > I'm not Randy, but I did state the previous definition.
    
    Yeah, I know, I was just particularly surprised by Randy's dismissive
    attitude.
    
        > B is utilizing A's transit resource in the manner
        > A intended.
    
    Purse snatchers utilize little old ladies' purses in the manner little
    old ladies intend?
    
    That's ascribing intention to A unreasonably.  You appear to be saying
    that if B is ill-intentioned enough to steal from A, that the fact of
    A's existence legitimizes B's theft.  Obviously, by this reasoning, A
    would not exist if A didn't _intend_ to get ripped off by B.  I find
    this argument unconvincing.
    
        > C is compensated by A to provide flow from B<->C<->A.
    
    Again, I'd disagree.  If A is buying transit from C, I'd suggest that
    A's primary intention would be to utilize that transit to reach point
    to which A is not already connected.
    
                 +---+         +---+
                 | A +---------+ B |
                 +---+         +---+
                      \  XP1  /
                       \     /
                        +---+
                        | C |
                        +---+
                       /     \
                      /  XP2  \
                 +---+         +---+
                 | D +---------+ E |
                 +---+         +---+
                 
    For instance, I'd argue that A would purchase transit
    from C if A wished to exchange traffic with D and E, on XP2, to which
    A isn't connected.
    
    For A to reply to B's traffic through C is an expensive waste of A and
    C's resources, not to mention a reduction in both speed and
    reliability for both A and C's customers.  Equally important is the
    fact that neither A nor C have any way means of gaining recompense
    from B for the expense and degredation of service which B is forcing
    upon them.
    
    Assuming A, B, and C are competitors, it's obviously to B's advantage
    to practice exactly this sort of idiotic scheme, since it has no
    immediate economic impact on B, yet the more B does it, the greater
    the expense to A and C.
    
    All the while, _all_ of the end users are experiencing a degredation
    of overall quality of service and increased monthly charges.
    
    So again, I ask how encouraging B to hose A and C's networks could be
    construed as building a better Internet.
    
        > C is rewarded for their compliance through an agreement with A.
    
    So what does that have to do with B, and how does it legitimize B's
    imposition at its sole discretion of an invented cost upon both A and
    C?
    
    I don't see a reward.  Just a stick which ISPs which place a higher
    value upon driving their competitors out of business than upon
    providing stable service can use to do just that.
    
        > It creates a better internet as A is encouraed to purchase QOS X
        > from C
    
    Excuse me?  How is the Internet improved by A's unnecessarily adding
    extra hops to both inbound and outbound traffic, passing higher costs
    through from C to its customers, and putting an unnecessary load on
    C's network?
    
        > A wants to talk to B more than B wants to talk to A.
    
    That's your original example, and the opposite of mine.  However, I'd
    say that that makes A a better ISP than B.  Your argument leads to the
    conclusion that the ISPs who try to provide the least connectivity
    should be rewarded most highly, through highest profit margins.
    
    That seems to me to be a socially undesirable end.
  
                         -Bill 

________________________________________________________________________________
bill woodcock  woody () zocalo net  woody () applelink apple com  user () host domain com
- - - - - - - - - - - - - - - - -