Lies, damned lies and product marketing

Greg Ferro’s “Layer-3 routing” post successfully kicked my huge sore spot: the numerous ways technical terminology is abused by product marketing gurus.

Twenty years ago, before networking became a multi-billion dollar industry, things were clear, simple and consistent: layer-2 (data-link layer) frame forwarding was bridging and layer-3 (network layer) packet forwarding was routing. Everything was crystal clear until some overly smart people tried to turn bridges into something they were not: WAN extension devices. A few large WAN networks were built with bridges … and failed spectacularly. Router vendors quickly used the opportunity to push the “routing is good, bridging is bad” mantra.

Fast-forward a few years. Routers were everywhere, “bridge” was a four-letter word and another group of smart people was trying to push large-scale LAN bridges. Obviously they couldn’t call them “bridges” anymore, so they appropriated a word from the WAN industry and started calling them “switches”. Thus “switching” (as in “LAN switching” formerly known as “bridging”) was born. The repainting of the old concept didn’t help much; they had to lure the customers with something sweeter. The cost difference between a simple layer-2 forwarding device and a complex layer-3 forwarding device looked promising, so the new mantra was “routers are too expensive, use switches”.

Fast-forward to late 1990’s. ASICs were getting powerful enough to implement decently fast and cheap basic routing functionality. It was possible to build simple cost-effective high-speed routers … only “router” was now a four-letter word. What could the product marketing people do? They invented “layer-3 switches”, which were really routers, but the new invention sounded so much better.

By then, everyone was thoroughly confused (primarily the customers … but that was also the goal of the whole exercise, wasn’t it) and extra verbs were being thrown into the soup, including “forwarding” which meant something very similar to what “switching” was supposed to mean before it’s been abused.

Today, you might see various interpretations of the same terms. I’m close enough to dinosaurs that I’d prefer to see “bridging” and “routing “ being used in their original context, but I guess we’re far beyond the point of no return. Failing that, using “routing” to mean the control-plane function (collecting, distributing and evaluating the reachability information) and “switching” or “forwarding” to describe the data-plane function is not a bad idea. But please try to stay honest: you always have to specify the OSI (or TCP/IP) layer on which the switching/forwarding activity is taking place.

17 comments:

  1. I love this post. May a times I have been zombied out by the stupid terminology..

    ReplyDelete
  2. Every time someone says "layer 3 switch" to me, I always say "you mean a router?". I just can't help myself!

    ReplyDelete
  3. We're too old 8-)

    ReplyDelete
  4. Interesting points. Nobody likes marketing but it is the way all business works. And yes, these ramblings do tend to make one seem old. Words are words, get over it. If a customer doesn't understand then they need to be brought up to speed or get out of the game...because they are too old/jaded.

    ReplyDelete
  5. In my experience, people mainly use switching because it's simpler. No ip plans, no strange dhcp server setups. Everyone can talk to everyone.

    ReplyDelete
  6. Constant cycle seen in small LANs:

    1) Build everything at layer 2 because "it's simpler".
    2) Scale a little.
    3) Things start breaking mysteriously. Run around in circles. Learn about packet sniffers and STP.
    4) Learn about layer 3 features in switches you already own. Start routing.
    5) Scale more.
    6) Things start breaking mysteriously. Learn about TCAMs. Start wishing for NetFlow.
    7) Redesign. Buy stuff.
    8) Scale more.
    9) VMWare jockeys start asking about bridging across the WAN.
    10) Enroll in hair loss program.

    ReplyDelete
  7. Yes you are ... ;-)
    I think nowadays it's really common that L3-switching is done on devices that do the forwarding purely in hardware (i.e. Catalyst), but routing is done without specialized ASICs ("normal" routers). And thats also what IT-managers are able to understand ... :-)

    ReplyDelete
  8. Fantastic 8-)

    ReplyDelete
  9. Nice post. I, too, correct people when they tell me that a switch is their default gateway or that their switch is running OSPF with their firewall. I tell them, then its a router. To which they inevitably reply, "no, its a Catalyst xxxx." MPLS adds to the confusion, but helps solve some of these problems. L2 across the WAN? OK. Segregated L3 across the WAN? OK.

    But anyway, Ivan and all.. Everytime some new technology starts coming down the line, it solves all of our problems. Remember LANE? It was going to herald a new era of video and collaboration! The Catalyst 8540 (The Cougar switch)... ahh yes... LANE. I'm so glad that died.

    Now PBT (or whatever they are calling provider ethernet) is going to save us. Bandwidth will be ubiquitous and cheap. We will have video and collaboration. And they will find a way to twist new meanings into old words as they roll it out with the same recycled marketing literature.

    Argh!

    ReplyDelete
  10. You don't work in product marketing, do you? O:-)

    The process is the same, once done in software, once in hardware. Both activities (regardless of their implementation) were called "routing" a decade ago. Not to mention that high-end routers have dedicated routing (aka L3 switching) ASICs and that most switches have to fall back to CPU-based switching once you exceed their hardware resources (or throw too complex packets at them).

    ReplyDelete
  11. Well, what bothers me is that some people create deliberate confusion and totally muddle the terminology to sell their stuff ... and that we all have to live with the results ;) But I guess you're right; it happens in all industries.

    I particularly liked the "get the customer out of the game" idea. Sometimes that might be very useful, but it's very rarely implementable.

    ReplyDelete
  12. You can also throw wireless in here as well!
    Gee.... how come I am not get 54 Megs on speedtest.net?

    ReplyDelete
  13. Imagine a room full of communicators, each with a series-connected switch to turn on the one light in the room. Now - how many people do you want in the room? And how do you communicate between rooms? ... Now I know I'm getting old ... Room == broadcast domain. Communication between rooms == router.
    Natch ...

    ReplyDelete
  14. I guess I should put my CCNA books I bought in 2006 on e-bay before they become "History of Networking" books.

    ReplyDelete
  15. Nizar: you just need to supply a word replacement list, such as L3 switch = router etc. and your book will be up to date again :-)

    ReplyDelete
  16. What do you mean by "But please try to stay honest: you always have to specify the OSI (or TCP/IP) layer on which the switching/forwarding activity is taking place."? Isn't switching/forwarding _always_ done at (OSI) layer 2?

    ReplyDelete
  17. Not since the routers were being called "L3 switches" ;)

    ReplyDelete

You don't have to log in to post a comment, but please do provide your real name/URL. Anonymous comments might get deleted.

Ivan Pepelnjak, CCIE#1354, is the chief technology advisor for NIL Data Communications. He has been designing and implementing large-scale data communications networks as well as teaching and writing books about advanced technologies since 1990. See his full profile, contact him or follow @ioshints on Twitter.