[c-nsp] ISPF

[Phil Mayers]

Anton Kapela wrote:
>> adjacent nodes.    Works the same way as an IGP, where you keep  
>> candidate paths around for faster convergence during a failure.
> 
> In the same vein, are folks finding ISPF stable in the field? Has anyone
> enabled this on their ospf speaking boxen in a "large" network?

Yes and yes, though the definition of "large" may be quibbled ;o)

> 
> Also, if anyone has done some lab trials I'd be interested to know the
> outcome. If folks haven't tested this (but could easily), perhaps they
> could construct an almost-full-mesh of a dozen or so routers, ISPF
> enabled on each, fill lsa db's with lots of type-2's, then script the
> oscillation of links within the network. Hopefully it wouldn't fail
> differently with ISPF disabled, but something tells me not to assume

I've certainly not noticed any correctness issues.