Provider credibility - does it matter? was Re:

===== Andy lague previously wrote: ====

>> So what's the problem? Basically, one organisation, Ebone, which
>> operates a fairly large pan-European net, won't peer with others. This
>> means that European customers connected to non-Ebone connected
>> providers like Global One lack European connectivity without going
>> through the US. Ebone provides great connectivity at cost, but the big
>> US NSPs making presences in Europe naturally don't want to buy
>> bandwidth from someone like Ebone. We'll see what happens.

Ebone is not the only way to get European connectivity without going to the
US. There are many exchange points in Europe providing local traffic
exchange. Global-One participates in some of these with more in the works.
Ebone is a good last resort (not to be read as default) path to obscure
destinations or to networks that will not peer with commercial networks.

Yes there are going to be more, many more peerings throughout the Europe,
and AsiaPacific, and Americas...

>I don't think this is correct. Ebone peers with other pan-european
>networks of comparable size, IBM, BT and EUnet comes to my mind.
>G1 is a customer of Ebone and connects in Paris.

Correct. Also Stockholm.

Maybe peering with EBONE is also a future possibility when we have more
common presence and comparable size. Andy is right, former Transpac
Sweden is now part of Global IP, and they also have a circuit to

Andy - speaking as a former Global-One employee

Thanks Andy. When is that lunch going to be? :slight_smile:



I've scanned this thread and I think most of the intelligent things
that can be said have been said. However, since Ebone was mentioned a
couple of times perhaps I should give one more comment:

Ebone is aware of the need for global connectivity and we divide the
world of ISPs into four:

1.Those we serve as customers and take payment from,

2.Those we do no-settlement peering with,

3.Those we pay to for connectivity and transit and

4.Those with whom we have no direct relation but to whom we connect
  via 1, 2 or 3.

For obvious reasons, we want to increase the number of ISPs in 1. and to
reduce or perhaps in the longer run eliminate the payment for category
3. But we don't have T3 lines to four or seven US NAPs at present.

In category 2 we try to put peers in the literal meaning of the word:
ISPs of a comparable size as measured in backbone bandwith and
geographical span. At this time, it looks like there are very few
such networks in Europe but that may change. Also, since our network
and other networks grow all the time, we don't have permanent and
exact criteria for when we peer but we do try to negotiate fair deals
with all potential peers.

There is no doubt that if we establish direct peerings with networks in
category 4 at one of our PoPs which to a large extent are co-located
with with the major European exchange points, then it would improve
service for both parties as often the hop count and delays would be
reduced. But its is difficult to offer no-settlement peering to an ISP
in, say Germany when we ask other German ISPs to be customers since
part of the Ebone offering is traffic over our international backbone
(the other part being transit). I think that as long as the cost of
international lines are much higher than the cost of national lines
you'll find ISPs with international backbones will be keen to
establish financially balanced deals on the use of those backbones.

More information on Ebone at