Here is a question.. a strange one, no less.
Is it feasible to do this:
WASHINGTON DC NEW YORK CITY
> B R I D G E |
MAE -- 100 Mb/s -- | Cisco | -DS3- | Cisco | -- FDDI or -- (multiple
East FDDI | 4700M | | 4700M | 100Base T peers)
giga Switch
Why? here's why.
Several folks in the same building in NYC want to connect to MAE-East.
But, we all don't want T1's or 10 Meg HLI to MAE-East, but DS3. So, this
allows us all to connect to the MAE, peer directly with others without an
intermediary ASN, and we can split the cost of the routers and the DS3.
I know (at least, I can't think of any reason it can't be done) that is
can be done. The unanswered questions are:
1) Will MFS allow us to connect multiple Peers on the same FDDI port (from
thier webpage, it looks like it, but I am not sure).
2) Is there any technical reason that the above is bad?
3) Because we do it the way shown above, does that make us look less
attractive (politically) ?
Thanks for any input on this. If there is anything I am missing, please
slap me. Thanks.
Technically, I believe that this is doable. Using FDDI in NYC would cause
the fewest technical issues. Note that if the DS3 is to be filled, it may
require more than a 4700.
Please note that this does not constitute an endorsement of the strategy or
of the 'taste' of such an operation. 
Tony
Here is a question.. a strange one, no less.
Is it feasible to do this:
WASHINGTON DC NEW YORK CITY
> B R I D G E |
MAE -- 100 Mb/s -- | Cisco | -DS3- | Cisco | -- FDDI or -- (multiple
East FDDI | 4700M | | 4700M | 100Base T peers)
giga Switch
You may want to go with 7500s, I think the 4700s will have a hard time
when the ds3 starts to fill up.
Why? here's why.
Several folks in the same building in NYC want to connect to MAE-East.
But, we all don't want T1's or 10 Meg HLI to MAE-East, but DS3. So, this
allows us all to connect to the MAE, peer directly with others without an
intermediary ASN, and we can split the cost of the routers and the DS3.
I know (at least, I can't think of any reason it can't be done) that is
can be done. The unanswered questions are:
1) Will MFS allow us to connect multiple Peers on the same FDDI port (from
thier webpage, it looks like it, but I am not sure).
They would 12 months ago, when I wanted to do it.
2) Is there any technical reason that the above is bad?
Strange, but I don't think it is "bad". When I needed my Ameritech NAP
connection up ASAP and did not have a space for it. I had Ameritech cross
connect my NAP DS3 to Wolrdcom and extended it to ATL. People thought it
was odd to have 20 ms delay to a NAP connection, but it worked.
3) Because we do it the way shown above, does that make us look less
attractive (politically) ?
Could be, I think it will depend on how you educate your users. If you get
a lot of people that are defaulting to someone or generally screwing
things up then yes. If you make sure they have a clue, then I don't think
it would be a big deal.
==>
==> WASHINGTON DC NEW YORK CITY
==>
==>
==> | B R I D G E |
==>MAE -- 100 Mb/s -- | Cisco | -DS3- | Cisco | -- FDDI or -- (multiple
==>East FDDI | 4700M | | 4700M | 100Base T peers)
==>giga Switch
==>
Yucky.
==>1) Will MFS allow us to connect multiple Peers on the same FDDI port (from
==>thier webpage, it looks like it, but I am not sure).
The more germane question here is "will MFS allow us to extend layer 2
across all these devices to provide a multi-access point in NYC?".
==>2) Is there any technical reason that the above is bad?
Bridging is signficantly more troublesome to troubleshoot. Additionally,
if these providers love layer 2 so much that they connect their "MAE port"
into another switch, and have other interconnects using a bridged
environment (yes, I've seen it), getting all the parties to cooperate in
debugging spanning tree problems can be difficult.
Additionally, if the DS3 between Ciscos is done on an ATM card, keep in
mind that you'll lose close to 30-35% of your traffic because of the ATM
cell tax--you'll probably get a max of 30 Mbps throughput. If it's
HSSI/HDLC, you shouldn't have many problems.
Also, while translational bridging has been around for a while, you may
experience problems from FDDI->FE.
==>3) Because we do it the way shown above, does that make us look less
==>attractive (politically) ?
It's a creative solution, I'll give you points for that.
/cah
Also, while translational bridging has been around for a while, you may
experience problems from FDDI->FE.
Which brings to mind... does anyone have a pointer for a two port,
FDDI<==>FE translational bridge? Less expensive is better... 
Yes, I know I could do it on a Linux box, but let's just not go there, OK?
---rob