The cost of change...

Ouch! I was just reading the article "Cisco phases out 1700, 2600 and 3700 series routers" by Phil Hochmuth at Network World. It seems Cisco recently announced that it has marked the 1700, 2600, and 3700 series routers for end-of-life. Can you say "costly upgrade?" I thought you could.

I'm often asked by press, analysts, potential customers, etc., about the cost difference between a Vyatta OFR and a closed-source soluition. People seem to understand that open source provides an up-front cost savings in terms of using commodity hardware rather than proprietary hardware, but people frequently miss the costs associated with ongoing change.

If you're an organization that just bought a Cisco 3700 series router last year, you will probably be upset today. I would be. See, after March 2007, you won't be able to buy a Cisco 3700. If you want to do something like upgrade the system after March 2007, you'll be forced to scavenge your parts from the used market (eBay, et. al.). And we know that Cisco looks down on anything purchased from the used market. In March 2012, you won't be able to get Cisco to support you at all. That sounds like a long time from now, but last week Cisco just relegated your $8,000 to $12,000 box (base price, more with options and interfaces) to a status just above that of a paperweight. Read the article for some descriptions of the organizations that are having to buy all-new replacement 3800 routers today. Also note that a 3800 is more expensive ($9,500 minimum). If you have a large deployment, Cisco just forced you into a large purchase.

Note that if you had purchased an open source product you probably would have saved more than half right off the cost of the initial purchase, and even if your particular hardware model had gone obsolete in a couple years, you'd still be able to buy all sorts of things to plug into it for quite some time. Things like individual x86 processor models and even PCI busses may go by the wayside, but the Vyatta OFR will continue to support those things for a very long time indeed (as long as the Linux kernel does) and you have the whole ecosystem of commodity hardware vendors from whom to purchase additional products in the mean time. Thus, if a PCI Ethernet card fails 7 years from now, you'll probably still be able to buy a replacement from somewhere.

Note that Cisco isn't the only bad-guy here. All the closed networking companies obsolete hardware from time to time. The Network World article is just a timely example. As a product manager working for some of those companies, I'm very familiar with the EOL drill and with the costs to the company of supporting old hardware models. The benefit of an open platform to users is that a whole ecosystem exists to support all the various pieces of gear. Even if a single vendor stops shipping something, you can much more easily transition to something from another supplier. You aren't held hostage.

So, what's the price difference between a closed solution versus an open one? For many organizations right now, it's at least $9,500 times the number of new 3800 routers they need to buy.