On 25 Nov 2003, at 14:36, Juha Saarinen wrote:
I'm pretty sure that the vast majority of TCL
customers, yours truly
included, questioned the wisdom of said business decision that
effectively locked them out of sites that in some cases they needed to
Maybe. On the other hand, perhaps the majority of TCL customers
probably didn't notice, and if they did they ascribed the problem to
the sites they couldn't access rather than TCL. Look! I can reach
Google, but I can't reach IDG. IDG must be broken.
In this scenario, maybe it's ICONZ that loses customers, not TCL. Or
maybe ICONZ pays money to restore access to TCL in order to maintain
the level of service that allows them to retain customers. That's
presumably what TCL is hoping; it sounds like the same kind of
de-peering game played by UUNET and PSI and C&W and all kinds of other
people at various times in other places (and lo, the Internet
It strikes me that Clear and its techies (hi Joe!) had
different idea as to the benefits of peering than Telstra has
I think the geek crew at TCL today probably have about the same
opinions as to what makes a good network good as we had back when we
had our hands in the routers in AS4768. Path diversity and connectivity
are good, and the more you have, the more control you have over the
network traversed between your customers and other peoples' customers.
When your packets travel over other peoples' nasty bits of mouldy old
cable, they're out of your control, and when your customer complains
about performance, there's nothing you can do about it.
Did the average customer notice or care that domestic paths were more
reliable or better through 4768 than they were through other ASes?
Probably not. (Maybe they weren't better; I have a somewhat blinkered
I suspect that any de-peering that is going on at TCL is based on
business decisions which are not made by the geek crew. I don't think
it's reasonable to attribute this stuff to those people.
Certainly, the APE Web pages seem to say peering is
beneficial rather than a commercial resource to be milked, but maybe
I'm showing my ignorance ...
Perhaps the issue isn't whether peering is mutually beneficial but
rather whether de-peering is mutually painful. If there's an inbalance
of pain, there might be a competitive advantage that arises from doing