[c-nsp] Inverse multiplexing DS3s to larger interfaces?

[Thomas D.Simes]

Thanks for the detailed reply Hyun, I agree 100% with your analysis and
order of preference from a technical standpoint.  The question of
whether the carriers will give us OC-X is more of a regulatory question
unfortunately and we are doing all we can to pursue that path.


On Wed, 23 Feb 2005 15:25:00 -0600
Hyunseog Ryu <r.hyunseog at ieee.org> wrote:

> If you want to avoid to hit maximum-path 8 limitation,
> you have to increase the pipe size, and reduce the number of circuit 
> from the router viewpoint.
> Ideal solution is to replace DS3s with higher bandwidth circuit such
> as OC3, OC12 or GE.
> But as you said, there is limitation from one of your provider, who 
> don't support more than DS3 pipe.
> 
> In that case, I can think of a couple of ways.
> 
> 1) Try to find another carrier who support OCx level circuit, or 
> pretending to do that.
> 
> If you are lucky, you can find another provider, who can fill up OCx 
> requirement, and your headache may be gone.
> Or you can use it to threaten current provider to support OCx circuit.
> If they have multiple DS3s, I don't see the reason why they couldn't 
> support OCx circuit.
> If they don't want to spend the money on OC3c cards, it's their
> problem. In order to support multiple DS3s, they should already have
> OCx support capability.
> 
> I think this is best case for you, because you don't have to worry
> about 
>    TDM Mux/demux between different speed, and routing may become
>    simpler.
> 
> 2) Find the vendor who have DS3s-to-OCx converters.
> 
> Kentrox or Larscom, or others.
> They may have some converters to map multiple DS3s to single OC3c or
> OC12c. This may be cheaper solution, and you can buy OC3c or OC12c
> interface for GSR, and keep DS3s circuits as-is.
> Routing may become simpler, but if there is some difference in mapping
> 
> speed from either side, there may be some issue.
> For an example, if you have 2 DS3s, and you map 2 DS3s to 1 OC3c,
> there is difference in capacity from both side.
> Actual circuit side is 2 DS3s, which is 88.420Mbps.
> Router side is 1 OC3c, which is 155Mbps.
> So if the router send the traffic more than 88Mbps or there is some
> problem with TDM multiplexing, there may be packet loss or packet
> transmission delay.
> 
> 3) If you want more sophiscated equipment, you can use Cisco ONS 15454
> for 2) function.
> 
> For the quality of transmission, and long-term management,
> if I were you, I will try solution 1) with some scalability planned.
> Maintain 2 carrier providers, but if they couldn't meet your 
> requirement, don't be afraid to ask for yourself to find another
> provider to suit your needs.
> 
> If they can sell x number of DS3s, there is no reason not to support
> OCx for equivalent bandwidth. It may take a couple of months to
> install fiber, but it's not a problem you can wait.
> 
> 
> Hyun
> 
> 
> 
> Thomas D.Simes wrote:
> > We've currently got a pair of GSRs tied together with 7 DS3s from
> > two different carriers and are load balancing via OSPF.  This
> > method has worked pretty well, but the technique tops out at
> > maximum-paths 8 on the GSR.
> > 
> > I'm looking for a more scalable way to multiplex these DS3s into
> > something larger and feed them into my GSRs to reduce the number of
> > layer 3 paths between the two locations.
> > 
> > Going to larger interfaces is currently not an option with one of
> > the carriers, and we want to maintain dual carriers for redundancy
> > if possible.  We're currently using the 6 Port Packet over DS3 cards
> > and the interfaces don't support multilink PPP or multilink
> > frame-relay.
> > 
> > What have folks used that worked well for this application?
> > 
> > The Cisco ONS 15454 using enhanced DS3 cards looks like it
> > might be a possibility, but I don't have any experience with the
> > platform.
> > 
> > TIA!
> > 
> 
> 


-- 
Tom
 
======================================================================
   "Z-80 system stack overflow.  Shut 'er down Scotty, the system's
         sucking mud" - Error message on TRS 80 Model-16B

Thomas D. Simes                                 simestd at netexpress.com 
======================================================================