Mailing List Archive

> The net's usability slowly declined over the years until reaching a
> level even network engineers couldn't hand over, then it would
> suddenly take a leap to a new bandwidth plateau.

> How about a few positive suggestions which could be implemented in
> the short-term.

Sean:

Get more bandwidth.

--jon.

Dennis:

Right. We are way overdue for a major step up in bandwidth. What we
need is OC12 bit pipes (no stinking ATM cells) and Routers that can
handle them.

--jon.

>>>>> "Jon" == Jon Postel <postel@ISI.EDU> writes:

Jon> Sean:

Jon> Get more bandwidth.

Yup. Get your chequebook ready.

Sean.

Jon,

what you suggest would be trivially obvious except that it isn't quite
right.

by the time we get anything that new installed, it had better be able
to go a lot faster than OC12 or we'll be right back in the same pot
on the day we put it in service.

Native packets over SONET is indeed The Right Thing To Do.

The more interesting question is what is the difference between
switching and routing, and what is the role of each as the networks
grow in size, capacity, and bitrate. (And no, I don't mean cells.)

It is fair to say there are both technical and religous positions on
this topic. (grin!)

-mo

> From: "Mike O'Dell" <mo@uunet.uu.net>
> by the time we get anything that new installed, it had better be able
> to go a lot faster than OC12 or we'll be right back in the same pot
> on the day we put it in service.
>
I disagree with this position. The fibers installed to do OC-12 will
certainly do OC-48 when it becomes available. And the next x4 jump as
well.


> Native packets over SONET is indeed The Right Thing To Do.
>
As author of PPP over SONET/SDH, I get queries from time to time about
when this will be available. There are at least 2 unannounced products
in development. I expect they will begin delivering OC-12 next summer,
and OC-48 the following year.


> The more interesting question is what is the difference between
> switching and routing, and what is the role of each as the networks
> grow in size, capacity, and bitrate. (And no, I don't mean cells.)
>
I also expect that _routing_ at that rate is the hard part. Maybe being
a link kind of guy, the link parts just look easier to me.

I expect _switching_ at that rate to be even harder than routing if they
are cells; the times are just too short. But the usual packet size
isn't much more than 3-6 cells, so the scale isn't as much better as we
would hope.

We tried to make 1500 bytes ubiquitous with PPP, but too much host
software still uses 512 bytes anytime it thinks the destination is
"remote", even when there is 1500 bytes end-to-end.

Bill.Simpson@um.cc.umich.edu
Key fingerprint = 2E 07 23 03 C5 62 70 D3 59 B1 4F 5E 1D C2 C1 A2

Bill,

I never said anything about switching cells. Why would I have
mentioned native-mode packets over SONET if I intended to switch
cells??? (grin)

Yes, the fibre can go to OC48, but if the boxes stall out again
at OC12, by the time we get a new generation installed, it could
be looking at a serious lifetime limit.

-mo