Need help from UUNet's routing specialist!

For the record, is anyone seeing even vaguely decent connectivity via
teleglobe? It seems that 50% of the times customers complain they have
a poor connection to our network, it turns out that they're going via
teleglobe at some point!

Um, maybe it has to do with your connection to the outside world?
(BTW, are you refering to, specifically, or some other network?)

Or at least connectivity between your upstream provider and either you or us...
Currently we hear your routes only from insnet, via LINX and DE-CIX.
(LINX does tend to have problems from time to time.)

The proportion of complaints coming from customers of ours is most likely
explained by the size of our customer base; a recent check on percentage
of announced IPv4 space, shows us and our customers comprising about 20%
of reachable addresses (ie, pro-rated by prefix size).

Are they running a 2400baud modem with a flaky fuse as their main
connection or something?

Try multiple OC-12 (cross-section), with buildout to OC-48 (on dark fibre)
underway, in North America and Europe, and deployment of significant numbers
of nodes worldwide at significant connection speeds, over owned wet-fibre
and dark-fibre capacity.

We do have lots of *customers* whose links to us are congested; if anyone
can find a way to force customers to upgrade, we'd dearly like to know. :wink:

(These congested links have in-addr entries from our namespace, and this
frequently leads to the incorrect conclusion that it is our network that
has problems. In case anyone is curious, our naming convention for point
to point links, is if-* for the Teleglobe end, and ix-* for the customer
side, where the * is the numbering of Cisco interfaces converted from
non-alpha to hyphens, in the zone named by the router designation.
Example:, would be the Teleglobe side,
while would be the far-end of a link.)