[c-nsp] Your opinions on router throughput

[David Freedman]

> The PA-A3-T3 cards are
> on a point-to-point DS3 and were used instead of regular serial DS3
> cards because they had them around.
> 

You will also have to bear in mind the ATM overhead which will reduce 
the effective throughput of this solution

See this old nanog post for more information (assume that PLCP is 
enabled by default)

http://www.merit.edu/mail.archives/nanog/1997-03/msg00263.html

I dont think much has changed here, unless the PA-A3 card has a way of 
doing this better....


Dave.


> Both VXR's have NPE300 with 256MB of ram.  Both are running
> iBGP between themselves, and eBGP to the feeds and receiving a full
> BGP table from the feeds.  IOS is 12.2
> 
> Right now everything is happy and there's no problems.  But, utilization
> on the feeds is nowhere near max.
> 
> The question put to us was,
> 
> Is there a chance of the NPE300's running out of horsepower
> as throughput on the setup increases to the maximum of each circuit?  There
> are customers on the LAN side that could, if all bandwidth limiting
> was removed, saturate either the 10Mbt-burst-to-45Mbt, or the 30Mbt ethernet
> feed, or the 45Mbt interconnect between routers.
> 
> I told them to try it and see what happens. :-)  That didn't go over
> very well.
> 
> My gut is that this setup, while it may run utilization up past 60%
> at full bore, will pass muster.  Certainly the chassis of the 2 routers
> will, but I am unsure of the real-world expectations of an NPE300.
> (as opposed to the published specs)
> 
> They are running ip cef.  They know that the NPE300 is
> obsolete but there is nothing in the newer IOS that they want.
> 
> Comments and speculations welcome.
> 
> Ted
> 
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>