We all have things we love and hate about our job. Sometimes it’s silly process stuff like TPS reports…but many times the good far out weight the bad. Just like picking out who to go out fishing with. Folks are gonna get on your nervous and make you want to hit them with the boat paddle, you just pick the folks that make you wanna do that less than others. I think that was my wife’s strategy for picking out a husband too. Not to say that I haven’t been clocked with a boat paddle, skillet, metal detector, etc…
<Long pause of reflection> Anyway…
Cisco has many cool technologies for sure. And by technologies, I mean features. When I was competing against Cisco way back in the dark ages when we pushed packets up baluns…both ways! if a customer said they are going with Cisco for one of these features I cut our losses and headed for the next customer. This is a blog of my three favorite Cisco features that only Cisco has and is a great reason to buy Cisco besides the neat-o blue console cables.
Cool Feature 00×01: ISSU (In Service Software Upgrade)
Engineer’s dilemma; How Do I upgrade my switch code when my network is 24x7x365 with minimal to ZERO downtime in a five nines world? When I FIRST came over to Cisco as an employee after getting my badge (and having a burrito at the Cisco café in building 17), I had to talk to the folks that developed ISSU. At my former gig we tried all kinda of things to get this feature to work on our gear. We just ran into tons and tons of problems. At first we tried modifying VRRP with something called XRRP and it just sucked. Customers hated it. Then we tried other snapshotting technologies and just couldn’t get close. See the problem is how do you GRACEFULLY start two primary software threads to the same process then keep increasing the NICE value so the secondary in now primary without panic-ing the kernel AND losing packets. Oh and remember I need a back out process if I abort or lose power, a roll back process if the new code starts causing problems, and a solid commit process to give me one more chance to back out so it’s not a permanent change until I agree. It’s a huge development challenge that took Cisco many many thousands of engineering hours to figure out. As a coder, I really admire the team that designed this technology and overcame all the obstacles to make this work. ISSU is dependent on NSF (Non Stop Forwarding) and SSO (Stateful Switch Over) and is hands down my single favorite feature at Cisco among all of products. If there was a Noble prize for networking; this would win hands down.
Cool Feature 00×02: STP (Spanning Tree Protocol)
STP is in nearly every switch out there today. However….with STP in its IEEE state you can really get in trouble quick. The Cisco difference is HUGE here. First off, STP is enabled by default. In an Auto-MDIX world, this is critical. I don’t know how many customers I walked into that have had ports cross connected by users. Sometimes, for fun and sometimes, well, I had one user tell me that they cross connected a port because the cable was a trip hazard…Yep… safety…ummm…first I reckon…. Now consider all the options Cisco gives ya to protect your network from easy hacks and misconfigs; Port Fast, Uplink Fast, Bridge Assurance, BPDU Guard, Port Guard, Root Guard, Loop Guard, Etherchannel Guard, all the filtering AND PVST simulation to allow MST to interop with Rapid PVST+. Nobody does STP better the Cisco. This is a deal killer if you’re a competitor. And, if you wanted the best description and write up on Spanning Tree protocol outside of the book; “Interconnections” by Radia Perlmann it’s hard to beat Cisco LAN Switching Fundamentals.
Cool Feature 00×03: OTV (Overlay Transport Virtualization)
When I was first induced to this feature, I was told it was “mac routing” ????WTF??? Mac….Routing?? Humph…Marketing people. I need to talk to the coders to see what this REALLY is. Well, their kinda right…it’s really more like a distributed control plane among all DC’s using the MAC address to advertise reachability. Designed for the data center interconnectivity it works by encapsulating MAC frames destined for remote DC’s in an IP packet, transports it across the network to the target DC where it unwraps it and forwards it on. OK, so how is that different from routing right? Here’s the thing, first, it doesn’t use the data plane for this. Honestly, (with respects to John Moy) to me, it’s more like L2 OSPF. Each Nexus head end switch keeps a MAC table for each remote destination and can recognize a MAC frame bound for another DC without having to climb the stack to do a routing look up. Heck it even handles multicast efficiently too. PLUS you can config and set this up in about 4 to 8 commands. Cisco likes to say four (marketing pays the bills baby) my experience is about 8. I also like that broadcast storm containment is built in so we don’t flood rogue frames out to all DC’s and it has loop prevention, multi-path, a rudimentary form of load balancing (layer two right?). My fav feature inside of a feature is the fact that we can add a new DC onto the OTV diameter and only config the new DC. The others will automagically update their tables to include the new DC to the OTV par-ty!! This is one excellent and well thought out feature. It’s everything TRiLL should have been.
If you could boil it down to three features, what would be your top three? List ‘um out and let’s see what ya think! It just goes to reason that there is a light and a dark side to the Force. Tune in next week to see my three Cisco features that I really dislike and why. Sure to piss some folks off without a doubt!!
Jimmy Ray Purser
Trivia File Transfer Protocol
-40 degrees Fahrenheit and -40 degrees Celsius are the same temp. It’s the point when both temperature scales converge and everyone agrees that wholly friggen smokes it’s too cold to be measuring anything out here, let’s get our tail back inside.