Re: [RAM] Number of DFZ routers

On Wed, 30 May 2007, Simon Leinen wrote:

> Chris L Morrow writes:
> > On Tue, 29 May 2007, Simon Leinen wrote:
> >> I think Robin's question was about *operational* routers, i.e.,
> >> configuration of routers in actual use in the Internet, not about
> >> hardware/software limits:
>
> > there was a reason I quoted the idb limits... on a SP network more
> > ports (logical and physical) you can stick in a rack the better you
> > are. SP's are always driving for more ports in less
> > face-plate-space. The IDB limit is probably a good start for today,
> > and tomorrow I'd expect SP's to drive that higher...
>
> Sure, we all LIKE to have large routers with many ports, preferably
> each generating LOT$ of revenue.  (And hopefully even *our* tiny
> routers will by upgraded over time to break Cisco's IDB limits :-)
>

I didn't say we'd like to have, I said that the IDB limit is there and
being pushed for a reason. It's not uncommon to have oc-12->ds1
channelized interfaces today at, 330+ ds1's per interface. It'd be not
uncommon to have 3-4 of these on a single chassis with 2 uplink
interfaces. That's 1200 interfaces on a single device.

>
> So Tony Li tells us a CRS-1 could support a million DS3 interfaces.
> But how many DS3 interfaces - or any kind of interfaces - were sold
> for CRS-1s? Let's try to put some real-world data into the discussion.
> What I'm hearing so far reminds me too much of "traffic doubles every
> 100 days!".  Not good for basing rational decisions on if you ask me.

Also keep in mind that mutli-chassis systems exist now, I'm not sure
(perhaps Tony can say?) if the IDB limits scale with multi-chassis or stay
constant for a single 'system'... So, more interfaces still per 'system'.
There are multi-chassis systems in existence today in the field even.

_______________________________________________
RAM mailing list
RAM@iab.org
https://www1.ietf.org/mailman/listinfo/ram