If the Swiss army is a Cisco customer, I’m pretty sure the Catalyst 6500 would have been their switch platform of choice, but it looks like the end of the road is in sight for this model in our networks. We’ve had 6500s since the beginning. The 6500 has been in the core of the data center, the aggregation layer for our WAN, and in all the wiring closets, but now each of the jobs can be done better, faster, and cheaper by newer products. (Newer Cisco products, so don’t dump your CSCO stock just yet.) Our new data center is going in with Nexus, but we’ve been moving away from the 6500 for a while.
Service modules were the first to go. The concept was great – we started with CSM in front of our servers for load balancing and health monitoring. Then when the application needed encrypted connections, in went the SSLM, and all was good. We had FWSM to segregate the zones, and even the odd NAM card to capture packets.
But the modules got old. NAM went, but instead of NAM2 we replaced them with standalone sniffers with significantly better capacity and features. The CSM and SSLM went EoL and were replaced with ACE appliances. FWSM soldiers on, but the limitations on performance, virtualisation and rule set sizes make ASA appliances more attractive. Plus, the security team wants to see vendor diversity in our firewalls, so some FWSMs will become Checkpoint appliances.
Service modules may have suited our requirements in the past, but now we prefer appliances. With a choice of models to suit different applications and standalone, appliances are easier to support. Hosting service modules inside the switch was good for reducing cabling, but had little else going for it. For example, software upgrades meant whole-switch and service module outages.
In the core, we upgraded the original Sup1 with Sup2 then Sup720, but where next? After its long gestation, the Sup2T is finally here, but it isn’t a straight swap for old processors. Pre-10Gbps Sup720 are end-of-life too, so we need to plan to replace them. The cost and disruption of moving to Sup2T (non-E chassis replacements, line card updates, PFC3 swapouts) is high, and makes a sorry case when compared to replacing the whole lot with Nexus. For the same amount of effort, you end up with an infrastructure that looks a lot more modern, which is why we went Nexus in the new data center.
That leaves our wiring closets. Moving to Sup2T would mean upgrading every 6513 chassis to 6513-E, and we’d need a refresh of line cards since the Sup2T is pretty fussy with older cards. For example, on the 6148A line card, only the newest model is supported. My thought is the access network would be better off moving to the 4500 platform. The subliminal message from Cisco seems to be that the 4500 is the preferred wiring closet platform.
The final nail in the 6500 coffin came from one of the storage engineers. He wants to move the corporate Windows shares onto a NAS with 10Gbps connectivity. How much to add some 10Gbps line cards to the corporate 6500 switches? More than the cost of simply buying a couple of Nexus 5Ks!
We ordered our last 6500 last year; we won’t be buying any more. Sad to say, this is the end of the road. Everything we do from now on will be replacing a 6500 in the network. The last 6500 will be laid to rest next to the last 7200 router that we should be taking out at around the same time.