Here Comes Open Source Telecom

My latest Business 2.0 article, The Black Box That Would Conquer Telecom, just went online over at the CNN Money website. This is a story about a stealthy startup called Vyatta, that has build the world’s first commercial open-source router, and how open source is slowly moving its way into the telecom world. Vyatta’s first product, an enterprise class router that will compete with Cisco-medium to low end offerings is currently in beta testing with some customers is based on XORP or extensible open router platform and runs off on two Intel chips.

The versatile open-source application can direct data traffic for a giant corporation as easily as it can manage a home Wi-Fi network. And that’s what makes it as disruptive as a leaf blower in a feather factory: Vyatta’s router will cost about a fifth the price of comparable models from big networking equipment makers such as Cisco Systems.

Vyatta is one of the many start-ups that are bringing open source disruption to the highly profitable and closed world of networking. While open source software movement has ravaged the bottom lines of companies like Sun Microsystems; networking behemoths like Cisco and Juniper have continued to enjoy fat margins they earned even before the telecom crash of 2000. Even today, a big portion of their IT budget goes into networking gear. Routers, switches, firewall devices, and even VPN boxes cost thousands of dollars.

“Open-source is providing real competition to the commercial telecom companies,” says John Todd, an open-source telephony expert. “It will force them to improve.”

The scramble for open source in networking comes because two primal forces tearing the old telecom order apart. First, the Internet-based technologies are replacing the closed legacy phone systems, thus helping the convergence of computer and the phone systems. In old times, in order to build a networking box, companies would design specialized chips, and run specialized software on them to get the best performance. Now you can buy extremely powerful processors like Advanced Micro Devices’ Opteron chips for a few hundred dollars, run special networking software on them, and get similar performance. There are nearly half-a-dozen open source projects that capitalize on the cheap processing power.

“I used to work in Novell’s multi protocol router group, but that failed because the chips were not fast enough,” Chris Ranch, Director of Network Architecture at data center operator Affinity Internet. “But you can do it all on a good PC.” His company is currently using open source load balancing software running on 15-pizza-box style servers that cost about $25,000. Similar gear from Cisco Systems or F5 Systems could have cost Affinity at least $750,000. “Given that we making money by selling hosting services, the cost of equipment is the difference between us making money or not,” says Ranch.

Corporations shopping for PBX systems are reaching same conclusions, and are turning to ultra-cheap boxes made by start-ups like Fonality, a Los Angeles company that packages open source Asterisk PBX software onto PCs running Linux. But no project is as audacious as Vyatta’s attempt to take on the highly lucrative and profitable router market.

Vyatta’s core brains come from XORP, a software router project started at ICSI in Berkeley back in January 2001. Atanu Ghosh a British-born researcher who works on the project points out that the software can be scaled down to run a simple enough home router on one end of the spectrum, to large-scale data network on the other extreme. “It is easy for third parties to extend the software, and I think people will come up with ideas to extend it,” says Ghosh.

The biggest interest in XORP and future Vyatta products will be in emerging economies like China and India, which are not cash rich, but have broadband ambitions. No one wants to pay for expensive commercial routers. “In the near future there would be ad-hoc networks on a person, and that could conceivably need a router with a tiny footprint, like XORP,” Ghosh predicts.