[c-nsp] Re: 7500 PPPoE dCEF aggregation

[Rodney Dunn]

On Tue, Jan 18, 2005 at 10:28:45AM -0500, Robert E.Seastrom wrote:
> 
> Rodney Dunn <rodunn at cisco.com> writes:
> 
> > There is a 2048 IDB limit on the 75xx and it
> > will most likely never be increased.  There
> > are multiple reasons why this box is not
> > recommended for broadband aggregation.
> > ...
> > If it were me looking for a platform to do
> > broadband aggregation I would not look at the
> > 75xx for this purpose.
> >
> > 72xx/G1, 7301, 10k are the most commonly used
> > boxes for this space that I have seen.
> 
> It's not as if the features aren't included, though as you mentioned,
> testing may be a bit spotty...  As with every other platform, it's a
> matter of calibrating one's expectations properly and understanding
> the limitations of the kit that you're rolling out.

100% agree from the scalability perspective.  It's when
something doesn't work technically that it gets more complicated.
If it works and can be used I'm for having it work in any
deployment scenario.

It's clear we are not there but in a perfect world if you
can configure it it should work and be fully supported by
the vendor up to certain scalability limits of course.
I'm talking more about the base functionality.
 
> 
> For example, I've had splendid success with 7505/rsp1 and 7010/rsp7k
> in proto-wimax proof-of-concept networks and in the lab -- the
> application is naturally low-bandwidth and in any event, 75-100
> customers would be a huge number to have connected at once.  For that
> niche, it's hard to beat the price/performance of above-cited boxes,
> even going with something like RouterOS on commodity Intel hardware
> (not supporting Radius attribute 242 (Ascend-Data-Filter) was
> decidedly a big minus, and 11 (Filter-Id) is not an acceptable
> substitute).
> 
> What this requires on the part of the technical team is a close
> working relationship with the financial folks.  It also requires trust
> between the two teams.  The basic proposition is "I can save you 90%
> on your startup equipment costs IFF you understand that this is NOT a
> solution 'to grow with', but rather a solution 'to get some revenue
> with', and that once there is revenue-generating traffic on the
> network, an upgrade of the equipment will be required in fairly short
> order, but at least your business model has been somewhat validated
> before you go spending big money on the real iron".

Can you put that on the web so I can use it as a reference? :)

Rodney


> 
>                                         ---Rob