Why I Hate Open Source Routing
Why is it that perfectly sane, intelligent, and bright journalists get boinked when it comes to open source technology? Mention Firefox and they interpret 10 percent marketshare as market dominance. They sing the praises of Asterisk not stopping to think...
March 1, 2006
Why is it that perfectly sane, intelligent, and bright journalists get boinked when it comes to open source technology? Mention Firefox and they interpret 10 percent marketshare as market dominance. They sing the praises of Asterisk not stopping to think that Asterisk's VoIP penetration is a pimple on Nortel's, Cisco's, or Avaya's you know what. The most recent example of "open source dementia" set in on Friday when Business 2.0's Om Malik broke the story about Vyatta, the startup developing enterprise routers for mid-size business and small-offices based on the XORP open source routing project.
Om speculated that the startup would undermine Cisco's routing sales. The company is using the XORP open source routing project to deliver routers for midsize businesses and remote offices, a $4 billion market, notes Om. Networking Pipeline's Preston Graila went even further to predict that XORP could lead to the demise of any vendor selling "high-margin networking hardware."
Now I've got tremendous respect for both Om and Preston. They're witty. They're funny. They've got a great handle on consumer VoIP. I also think XORP is a fantastic effort. Back at IT Architect (then Network Magazine) I helped write what was perhaps the first story on XORP.
But thinking that XORP will conquer the enterprise in our life time is pure poppycock. Business today is looking for more than a router. It's looking for networking architectures that involve routing, but also employ other technologies, such as security, switching wireless, or voice.
IT likes Cisco largely because it can present a soup-to-nuts solution that fits in with their existing environment. While that's true to the extreme with Cisco a similar argument can be made with other leading networking vendors, such as Juniper, Avaya, Nortel, or Checkpoint. Each of them can present holistic solutions that transcend a given product.Vyetta/XORP though won't be able to work in the language of routing that's spoken in today's enterprises, Cisco's IGRP. Without supporting the Cisco routing protocol, replacing Cisco isn't going to be happen any time soon. That's a big difference from LINUX, Firefox or other wunderkids of open source, where playing in the existing environment doesn't present nearly the same kinds of problems.And don't expect to argue that you'll just replace a Cisco core router with a XORP-equivalent. Besides IT's investment in the existing hardware, XORP's performance remains slower than Cisco routers because the forwarding plane is implemented in software. Routing code also has little do with the needs of the remote office, which today is looking for an office-in-a-box -- a router that provides all connectivity requirements. Cisco's Integrated Service Routers (ISRs) is a perfect example. It can be equipped with an IP PBX, firewall, a switch, wireless access point and an IDS/IPS. So far Cisco has shipped 750,000 ISRs.
Cisco isn't alone in that regard. Whether it's Juniper's J-Series or Adtran's NetVanta line, the message is one-stop delivery of the critical network applications needed in the branch office is what today's routing is all about.
The small business is another matter and I think here is where Vyatta/XORP's impact could really be felt. All sorts of small office, multifunctional gear will become more popular today because vendors can focus on the applications that really matter and less on the routing code. Alec Saunders makes this point when recounting a dinner conversation he had with a CTO of a very large PBX manufacturer. The CTO, blogs Alec, defined call control as a commodity and if he can outsource it to an open source project, it lets him assign resources to the far more lucrative applications market.
The other impact could well be in commoditizing Session Border Controller (SBC), notes Aswath Rao in his blog. He should know. He suggested the Inter Connect Function in the TIPHON Architecture that became known as the SBC. Rao writes that "in its core SBC is a 'twice NAT'. So it is very likely the open source router software used by Vyatta could be used to build SBC as well."Open source routing has its role to play in this market, but don't expect that to be in the enterprise any time soon.
About the Author
You May Also Like