RE: Core router bakeoff?

De : Neil J. McRae[]
R�pondre � : Neil J. McRae
Date�: mardi 12 mai 1998 16:39
Cc :; 'Andrew Bangs';;;
Objet : Re: Core router bakeoff?

> I use both CISCO GRF400 and GateD. I think that GateD is the best
> easy to use and develop public routing software. But even if the GRF400
> runs GateD, it's not as reliable as a CISCO 7500 per example.
> The GRF400 has to much bugs to be an operational backbone router.
> For example when you redistribute static routes via BGP, the GRF
> redistributes the adress IP of the non telecommunication port to.

This is rubbish and sounds like you haven't RTFMed. I redistribute
static routes throughout with the GRF and it works fine.

This problem happens only if you mention in gated.conf
  export proto bgp as xxxx {
    proto direct {
      all ;
    } ;
  } ;
Then the IP address of de0 is redistributed.

> The Ascend technical center is not helpful nor the documentation.
> And the performance are limited to an average of 50k pps per card.

I agree that the TAC is as good as useless the manuals of the GRF are
toilet paper, fortuently I've used BSD/OS and gated for a while so I
know my way around.

I've seen a GRF tested to more than 50000 pps so I'd like to hear how
you came to that conclusion.

I tested a GRF400 with Netcom Systems Smartbits with L3 cards.
Ethernet frames were 64 bytes long and the ethernet card of the GRF 400
began to drop packets when the aggregate througput reached 55000 pps.