Terminating many T1's

Wed Jul 23 12:41:53 UTC 1997

Hm, Neil, isn't the fact that we are in total agreement 
again one of the Signs of the Apocalypse?  :-)

"Neil J. McRae" <neil at domino.org> writes:

> The solution to this is for someone to make a
> Channelised SONET/SDH interface.

Designers should look at the Cisco (sorry, Neil)
CT-3 card: 

http://www.cisco.com/warp/public/733/CT3IP/ct3_ds.htm

which takes in an electrical PDH 45Mbps DS3 circuit and
pulls out 28 DS1s.  This is a beautiful card for anyone
aggregating lots of DS1s together, and I'm pleased Cisco
built it.  

(Consider that two CT3s get rid of most of the equipment seen
in this picture: http://www.stupi.se/Bilder/3266-2012-0599/jpg1/img0011.jpg)

Unfortunately it's all on the U.S. PDH standard, but hey,
what can you do? --:)

Like you, I would like to see one deal with SDH/SONET,
and be able to pull out European PDH formats as well
as U.S. PDH formats.  I believe there is a market for such
a card that consists of more than just you and me,
although big traditional telcos will still like the CT3
for a while to come (testing and procedures reasons mostly).

For the moment, however, if you have a telco supplying you
with U.S.-style PDH circuits over SONET, you can pull a
DS3 trib out of an OC3, and drop that onto the CT3 card,
and handle 28 DS1s that way.

Eliminating the need for the DS3 tributary card at the MUX
and pulling an optical connection into a router would be a
Good Thing as a next step, and I think that Cisco
understands this.

I dunno what your vendors are up to, but hey, if they want
to build such an interface, then that's way cool too.  The
end goals, just to amplify your comments, are:

	-- one pair of fibres from MUX to router
	-- many many many lines of various bandwidths
	   encapsulated in that pair of fibres
	-- de/aggregation of things like STM-1/OC3
           or faster bundles of PDH and SDH circuits
	   to be done by the telco side
	-- SDH/SONET is cool, and worth paying for
	-- SDH/SONET is cheaper facilities-wise for telcos
	   to provide

Ultimately I would like to be able to divide telcos into
two varieties: those that offer fast concatenated
point-to-point bandwidth (STM-1c to STM-16c, for customer
and inter-POP connections, for example), who will never
need to worry about PDH equipment ever; and those who will
aggregate large numbers of PDH circuits into SDH/SONET in
the standard way (to talk to lower-speed customers). [*]

This meshes nicely with the idea of evolving three
principal types of IP routers: ones that handle a number
of very fast POS ports (STM-16c is nice), ones that handle
one or two very fast POS ports and have massive
de/aggregation facilities, much like the Cisco CT3, only
at larger scale, and ones that are "special case" boxes
which trade off interface steroids for packet-per-second
processing time and general computational steroids (right
now these are generally what I would call "edge" boxes,
for use principally in talking with peers and/or customers
at exchange points where rate limiting and other policy
issues; in the future they could be huge NAT boxes, or
only heaven knows what :) ).

> SDH and SONET are multiplexing technolodgies and I wish the router
> manufacturers would wake up and grab this idea.

Agreed!  Talk to your vendors, I'll talk to mine. :-)

	Sean.

Footnote:
[*] For the sceptical, there are already alternative
    suppliers of high-speed concatenated SDH in Europe.
The traditional PNOs so far Just Don't Get It, and will
probably not take advantage of the potential revenue
stream for a while yet (big hint: if you're a potential
supplier of high-speed concatenated SDH and aren't
seriously thinking of pricing a few percentage points off
a linear extension of PTT tariffs, talk to me :) ).
The traditional PNOs are also no longer the only options
for offering up a bunch of PDH circuits concatenated into
SDH/SONET, either, so they face competition on that front
too.