Value of telephone numbers

[Karl Denninger]

On Mon, Jul 07, 1997 at 03:47:21PM -0400, Dana Hudes wrote:
> Karl Denninger wrote:
> > 
> > The generic problem isn't address space -- its router manufacturers
> > being
> > coddled by the Internet industry rather than being forced to produce
> > equipment
> > that can handle the real load out there.  Why is it that I can buy a
> > 100 MIPS
> > system -- a GENERAL PURPOSE system -- for well under $1,000 on the
> > secondary
> > market (this pretty much establishes the price on the CPU, by the way)
> > while
> > that same class of processor isn't at the core of any existing major
> > router
> > doing BGP computations?
> > 
> > Why is it that nobody is producing a router where there is a
> > *dedicated*
> > CPU that does NOTHING OTHER than recompute BGP tables?  (I don't count
> > prototypes, and yes, I am aware that there is *one* such device in
> > that
> > phase right now) Again, we're talking about something that isn't very
> > expensive to do from a design perspective, and would obviate many of
> > these discussions.
> > 
> 
> I don't recall whether the Netstar Gigarouter cum Ascend GRF has smart
> cards. I know it runs GateD.
> 
> Even if we have routers capable of running 150K routes, that costs money
> and it doesn't matter that its cheaper now to have a Big Fat Router
> than it was 8 years ago when the only thing around was the NSS at
> around $150K to run the NSFNET full T3 backbone. Its a competitive
> business enviornment out there and money spent on router memory is money
> that could be spent paying engineers. Or stockholders.

That's funny - - the ASCEND GRF is about half the cost of a comparably-equipped
CISCO 7513.

And yes, it does have smart cards and a Pentium main processor, runs a
BSDI Unix derivitive on the host processor with GATED for routing, and can
handle 150,000 routes (3x+ today's real load).

It works too (we've tested it under VERY heavy loads; I didn't care for their
ATM cards and have heard of problems with them, but the HSSI boards work
great, and the 100BaseTX interfaces -- great for local interconnect between
backbone routers in a data center -- are fantastic).

One has to wonder why it is that CISCO can get 2X the price for less than 
1/2 of the performance..... and why the industry has coddled a company which 
has produced products of this type.

My beef with CISCO on this matter goes back many years.  The blunt truth - 
their architecture has consistently failed to take advantage of the commodity
relationships within the high-tech sector, and the Internet industry has not
only condoned that failure, but allowed their engineers and employees to 
write many of the specifications which we now accept as "gospel" in the
community.

If this were any other industry, the fur would be flying.  Its only a matter
of time before it flies here, and the faster we wake up to this fact and fix
it, the better off we will be.

Ps: Anyone else hear a rumor about WCOM/UUNET doing a big deal with ASCEND?
Is this related to the GRF?  I'd be surprised if it wasn't.

--
-- 
Karl Denninger (karl at MCS.Net)| MCSNet - The Finest Internet Connectivity
http://www.mcs.net/~karl     | T1's from $600 monthly to FULL DS-3 Service
			     | 99 Analog numbers, 77 ISDN, http://www.mcs.net/
Voice: [+1 312 803-MCS1 x219]| NOW Serving 56kbps DIGITAL on our analog lines!
Fax:   [+1 312 803-4929]     | 2 FULL DS-3 Internet links; 400Mbps B/W Internal