While there is no difference from a technology perspective,
there's also no benefit to be gained by interconnecting large
networks at a public (as opposed to private) interconnects.
One can certainly make the argument that running large traffic
flows through shared interconnects is bad engineering if a
private interconnects for such traffic are available.
/John
>...
>But there isn't any gee-whiz technology that you can do at a
>private interconnect that you can't do at a NAP/MAE. Open NAPs
>aren't bad engineering.While there is no difference from a technology perspective,
there's also no benefit to be gained by interconnecting large
networks at a public (as opposed to private) interconnects.
John, you say there is no benefit to be gained by large providers using
anything other than private inter-connects. may i refine that question
slightly and ask: benefit **to whom?** to the large provider? If the
question of benefit is limited in this manner, i'd guess that your
statement is quite correct.
but what does this discussion begin to bring in focus regarding the
evolving topology of the internet? i wrote more than a year ago about the
"club of six" being at the top of the internet hierarchy by virtue of
interconnects and peering with each other at most of the five major
exchange points. Since then one of many obvious questions is: what does
it take for the six to become seven and the seven to become eight. Etc.
Since I do not have direct acces to routers at major exhange points, I
cannot easily tell who is peering with whom. not many people will talk
about this even privately. yet the conscensus has been that there will be
new members, because the alternative would mean inviting anti trust
action. Candidates are obviously companies like BBN, netcom,
advantis, crl, AT&T and others. Thus I hear statements like the club of 6
has now become the club of at least 15.... and presumably all is well.
When you get big enough, you to can connect to the internet at the very
apex as demarcated by major naps in the us and soon in asia and europe.
*BUT* here is my question. Don't private interconnects essentially
provide a new apex for the internet? One that pushes interconnects at the
major exchange points down a level. Sprint, MCI, UUNET, and BBN are
clearly the four largest players by market share. All of them have
multiple private interconnects with each other. perhaps ANS does as well.
AGIS certainly does not. So in this sense, the "6" have not really become
15 but rather have become 4 or maybe 5.
now you may say that from a competitive point of view this makes no
difference. perhaps. But what if the big four no longer see the need to
upgrade their bandwidth INTO and OUT OF exchange points? what happens to
the "secondary ten" when they get some large customers who see their
packects die between Sprints mae east router and the nearest sprint
backbone POP if that pipe is over crowded. Will we hear them complain
about ungodly packet loss and move to the industrial strength service of
the big four who can do hot potato hand offs to each other at multiple
private exchanges around the US and increasingly around the world? if
such is the case, how will the secondary ten ever get enough customers to
convince the top four to let them do private exchanges as well?
Is this part of an inevitable dynamic that is and will channel market
share into the hands of the top four? Given the need of the automotive
network exchange, will it, under these conditions, certify anyone except
the top four? One would certainly think it would certify the top 15 but
if the stratification between service at private interconnects and public
exchanges continues, could it certify anyone except the top four?
One can certainly make the argument that running large traffic
flows through shared interconnects is bad engineering if a
private interconnects for such traffic are available./John
True, but what about the economic and public policy aspects of such
decisions?
Gord',
You said:
"... But what if the big four no longer see the need to
upgrade their bandwidth INTO and OUT OF exchange points? what happens to
the "secondary ten" when they get some large customers who see their
packects die between Sprints mae east router and the nearest sprint
backbone POP if that pipe is over crowded. "
The argument can be made, and there might be empirical data to back it up,
that the private interconnects actually offload *at least in the short
term* the participants' pipes into the public exchange points. For
example, if a significant fraction (1/4th to 1/3rd) of the S and M and ...
traffic into a public exchange is S and M and ... talking among each other,
then offloading much of it at some other place might reduce the fractions
(maybe to 1/10th to 1/5th). Of course, if their business models do not
include eventual upgrading of their pipes into public exchanges, they will
have to balance the reduced <performance|reachability> to the Internet at
large as seen by their customers against any savings from not upgrading.
(Disclaimer: I pulled the numbers out of thin air, so anybody who quotes
them is as foolish as I for having put numbers in writing in the first
place.)
--Steve
Gordon Cook wrote:
now you may say that from a competitive point of view this makes no
difference. perhaps. But what if the big four no longer see the need to
upgrade their bandwidth INTO and OUT OF exchange points? what happens to
the "secondary ten" when they get some large customers who see their
packects die between Sprints mae east router and the nearest sprint
backbone POP if that pipe is over crowded. Will we hear them complain
about ungodly packet loss and move to the industrial strength service of
the big four who can do hot potato hand offs to each other at multiple
private exchanges around the US and increasingly around the world? if
such is the case, how will the secondary ten ever get enough customers to
convince the top four to let them do private exchanges as well?Is this part of an inevitable dynamic that is and will channel market
share into the hands of the top four?
Gordon - You're describing the dilemma of any newcomers to the net:
Assuming that the new net can get peering agreements at the public ix's
(this in itself is not easily assumed) there is still an uphill battle.
. If you don't have private interconnects, your traffic goes over the
90% avg. utilized links between the IX point and the large provider's
backbone. This makes it difficult to get and keep customers - after
all, 75% of the internet is lossy/slow to them, and if they switch to
any of the larger providers they don't see that loss.
. You can't get a private interconnect with another provider unless you
have the traffic (customers) to justify it. See previous point as to
why you can't get the customers.
Interesting points.
Rod
As one who is living this senerio on a daily basis, I can tell you it's
frustrating and upsetting. We have gone so far as to test the legality
of what is happening (there *must* be someone we can sue <grin>). Public
peering works well these days as the large networks move their traffic
off the NAPs, freeing up bandwidth for the mid to smaller networks.
The model that makes sense to me, is for the largest networks to exchange
traffic through private interconnects, and for them to treat the
aggregated NAP traffic as another large ISP. The NAP is then used for the
2nd tier and smaller providers to exchange traffic with each other, as
well as a collection point to gather up traffic for the large networks.
Unfortunately, at least so far, the first tier providers are not
supporting this approach, forcing me and others I assume, to purchase
connectivity.
Best Regards,
Robert Laughlin