[AusNOG] Cisco 6500 with Sup 720 3BXL - Good routing platform ??

[Matt Carter]

> > but as you already know, improving CL-TCAM behaviour was something we
> worked hard on with EARL8.
> 
> Concur w/this 100%!

As a person who has a large number of 65/76 in the field, I've been watching this thread both interest. I'm all for bashing Cisco where they deserve it ( just ask my SE ;) ) but I think the 65/76 claims are pretty extreme imho. I can empathise and appreciate where Roland is coming from, but it seems to be a very confined and narrow view, providing you aren't trying to jam a square peg into a round hole and understand the limitations, which I think is really what this boils down to, these boxes work just fine. 

Given the age of the silicon to say things like the netflow "just isn't up to snuff for production use" and "Hopefully, these edge-related shortcomings will be rectified by future hardware." may be all well and good, pointing to things like ASR's, which are nice and shiny, but for what 65/76 720+3b/3c are getting around for in the marketplace, bang for buck, I dispute they aren't "a useful edge platform on any network of any size" 

Yes there are limits, Yes there are caveats, No they aren't great for feature-rich edge operations, but for packets in/packets out, wholesale type interconnects, l2vpn, places where you already have other extensive mechanisms for addressing DDoS investigations shortfallings of this platform (which seems to be a focus of this thread), coupled with feature rich PE for the things you need it for, or managed CPE's doing end to end feature rich operations, leave these guys to do what they do best, they work just fine. 

"here in Australia ISPs have successfully used them for Netflow based billing for years" I don't know why anyone who has extensive experience locally would "seriously doubt that" ? I know a few people for whom a small 76 edge is/was merely a step up from the 72/73 to get a few hundred more mbps, particularly prior to the G2's, for what they cost, its a bridging solution that allows non feature rich traffic to be moved to such a platform, which can push out the point at which the likes of shiny new tin become a requirement, etc.

I just put these types of things in the category of platform quirks, there is a lot of great platforms out there, that achieve 95% of the aim, and have maybe one or two significant shortfalling, but does that mean we throw the baby out with the bathwater? There is nothing wrong with taking the good with the bad, provided you know what both the good and the bad are and have a plan for managing that !!

2c.

--matt