Internet II is coming...

"Dorian R. Kim" writes:

This sort of proposal, i.e. building a Higher Ed private network for
research, is in and of itself not such a bad thing.

The grow of Internet since NSFNet shut down has put serious strains on the
infrastructure that researchy folks used to use to do(and still do) their
various work on.

You know, maybe I'm crazy but I rarely see the troubles that people
mention so often.

When I'm going between my site and another site on the net, if both
ends are unloaded, I typically get bandwidth equal to the smaller of
the two pipes into the net. Its very rare that I don't get transfer
times near the maximum expected, even when one of the pipes is
attached to a mediocre provider. (Really bad providers are another
story, but I luckily can usually convince my clients not to use them).

Of course, if I'm communicating to a site that seven thousand other
people are talking to I'll get low performance, but between my clients
on new unloaded connections I rarely see any trouble. Every once in a
while someone's one router to somewhere critical melts down and
suddenly there is a massive bandwidth shortage over some provider's
backbone, but usually (call it at least 23 hours a day) everything is
just fine.

Now, I'm not satisfied with the overall reliability -- one hour a day
of not being able to get to an arbitrarily selected site on the net
doesn't seem good at all, especially compared with the phone system,
and especially if that one hour happens to be the hour that the new
economic statistics just were posted to the commerce department's FTP
site and people end up staying late waiting for the thing to get on
line so they can start the overnight batch calculations. However,
overall, things are pretty good -- I don't see these massive shortages
of bandwidth that people are talking about.

Seems to me that if the university researchers are sick of competing
with the undergrads, either the university could get a fatter pipe, or
they could priority queue the traffic from the researchers, and either
way they would probably win. Even with all the well-publicized growing
pains at the providers, I think the trouble is most likely at the end
points, and not in the providers.

Am I crazy? Are other people seeing massive bandwidth shortages that I
just haven't noticed? (There are some of these occassionally for a
week or two on some provider, but they rarely seem to last long.)


There is a "perception" of lack of bandwidth. MAE EAST is running at %30
capacity of a 100Mb switched FDDI. Figure there are about 30 providers
each with 45Mb pipe - a bandwidth shortage does not add-up - There are
many other metrics that effect performance besides bandwidth. How does
Internet II solve the other sources of performance problems?

That's Perry's view from New York City, now my view from Vernon, British
Columbia, Canada. This is a town of 40,000 population in the mountains,
6 hours drive from Vancouver, BC and 7 hours drive from Calgary, Alberta
which are the nearest two major cities.

My ISP has a 10Mbps fibre ATM circuit to BCNet in Vancouver that was
installed in Aplril 1995. From there it connects to CA*Net which has a T3
into MCI Seattle as well as links to the East where more T3's head south
to MCI. The T3's were T1's until Sept 1995 at which point there was a
small improvement in speed at times.

But I have to agree 100% with Perry. Speeds are great most of the time.
Congestion is generally something caused "out there" often at WWW server
sites. In my case the ISP has ample bandwidth to their provider and the
next two upstream links in the chain do a good job. IMHO this is the
reason I have such good service.

That's why I think most problems that people complain about are due to the
ISP's network. In the case of a university, they themselves are the
problem. This assumes a model of NSP connected to some sort of regional or
large ISP connected to your provider. The NSP's are good, the regionals
and large ISP's are usually good and the provider is usually the source of
the problems with bandwidth and congestion.

But these problems *ARE* fixable.

Michael Dillon - ISP & Internet Consulting
Memra Software Inc. - Fax: +1-604-546-3049 - E-mail:

I suspect that they will be going completely switched and have very few
total points on the whole network (depending on what configuration they
go to). A network of OC12s or OC48s in a redundant star will have
significant performance benefits because 1) no routing, or very symmetric
routing. 2) very low latency <8ms coast to coast I'd suspect. 3) priority
queues, quality of service, reserved bandwidth, etc.

They could also use the BFR from Cisco whenever that comes out, which is
supposed to do OC48 speeds :slight_smile:

I haven't read the Inet-2 information, so I don't know what kind of trunk
lines they are using, but I am pretty sure they want interconnect speeds
far faster than even multiple DS3 connects to each other, and they will
tolerate whatever internet-1 performance they can get.

I am not even touching the Mae-East at 30% fantasy. All I know is that the
UUNet <--> Sprint OC3 private connect at Tysons Corner is at better than
24Mbits average and mostly limited to router CPU problems.


Isn't this the reason that MAE-East is at 30%, i.e. there are now many
private interconnects between tier 1 NSP's to offload traffic from major
exchanges like MAE-East?

Not to mention that MAE-East is no longer the only major interconnect, a
fact that seems to be taking some time to work it's way into

Michael Dillon - ISP & Internet Consulting
Memra Software Inc. - Fax: +1-604-546-3049 - E-mail:

Just out of curiosity, how did you speed up the speed of light?




I am not even touching the Mae-East at 30% fantasy. All I know is that the
UUNet <--> Sprint OC3 private connect at Tysons Corner is at better than


24Mbits average and mostly limited to router CPU problems.

It's a DS3 at 29 Mbps ( as of a few seconds ago).
FWIW, our DS3s to MAE-E are at about the same load.
Clearly, private peerings have diverted a lot of load away from MAE-E
and other public exchange points.