There are few happy endings for enterprises that attempt to implement open source voice over IP (VoIP) without enough experience working with Linux and IP telephony systems. According to loyal users and software developers, successful open source VoIP deployments hinge on having that specialized expertise in-house -- or at least knowing when to suck it up and pay up for off-the-shelf systems that handle most of the configuration.
The flexibility and rich features of an IP telephony system based on Asterisk, the most popular open source VoIP project, made retiring a 26-year-old Comdial Corp. public branch exchange (PBX) in 2008 a lot easier on Dan Peck, vice president of Diamond Tour Golf, an online retailer with about 25 users and headquartered in DeKalb, Ill.
Commercial IP telephony systems didn't have the call recording and reporting capabilities Peck had found in open source VoIP software; commercial systems also lacked the ability to easily change dial plans on the fly, he said.
But Diamond Golf only had one person who could run an Asterisk-based system -- its IT director, who had no experience using open source VoIP and a slew of other priorities to juggle, Peck said.
Instead of chancing it, Peck worked with systems integrator EUS Corp. to implement Switchvox -- a hardware and software system that minimizes the heavy coding and configuration for open source VoIP deployments. Sold by Digium, a private company that sponsors Asterisk, Switchvox is aimed at small and medium businesses (SMBs).
"I didn't have anyone in this building capable of saying that we knew more about phone systems than [our reseller] did," Peck said. "We got out of our own way in terms of the open source end of it and went with the simpler version. It suits what we need so we don't wake up one day and say, 'How do our phones work?' Understanding your limitations is one part of [a successful open source VoIP deployment]."
Without using a turnkey product such as Switchvox, a successful open source VoIP deployment requires an IT pro to be comfortable with IP networking, Linux administration, programming and traditional telephony, according to Steve Sokol, director of Asterisk Advocacy for Digium.
"Looping, executing functions, programmatic flow control, managing variables -- if you don't have those four basic skills, you're going to have an uphill battle with an open source project like Asterisk," Sokol said "In that case, you may want to look at something that's based on open source but a little more polished."
Finding the middle ground with open source VoIP
Open source and Linux weren't foreign for Jason Chase, technology director and head of computer science at Gould Academy, a private boarding school based in Bethel, Maine.
After maxing out the available licenses on his legacy Nortel Meridien PBX, Chase used Fonality Trixbox, a software PBX based on Asterisk, to connect a dozen users in a new branch office. He later extended Trixbox to deploy IP telephony in 50 classrooms that had Ethernet jacks but no analog lines.
When it came time to replace Gould's 15-year-old legacy phone system on a $28,000 budget last year, Chase returned to Asterisk after commercial vendors estimated $90,000 to $200,000 for a full implementation.
The open source VoIP system cost the academy $20,000, he said. The upgrade replaced his legacy Nortel desk phones and PBX with Aastra IP phones, a Dell PowerEdge server and a support license for AsteriskNOW, an Asterisk-based software package from Digium designed to simplify custom telephony application development but offer more control than Switchvox.
"Since it's an open source piece of software, any feature that's in the system is available to us. With the Mytel and Cisco solutions, every feature we wanted to use was an additional licensing cost per user," Chase said. "We've used a lot of features in the Asterisk system that I don't think we would've gone out of our way to pay for in other systems."
Open source VoIP: For SMBs only?
Open source VoIP still has limited adoption, according to Irwin Lazar, vice president of communications and collaboration research at Nemertes Research. The widest adoption comes from technology companies and universities, where end users are likely to lend support, he said.
"If you just want the cheapest possible phone system for 30 to 40 people … it's hard to beat Asterisk," Lazar said. "But if you're trying to do something more complex -- like trying to integrate Microsoft Office server or a videoconferencing server -- it's hard to do."
Larger enterprises are less likely to try open source VoIP and tend to stick to the brands they know, he said.
"The challenge we've seen is scaling up to larger companies. There aren't that many organizations that provide professional services around it, and it's not something that's as easy to deploy as Cisco and Avaya," Lazar said. "Cost savings become a little more negligible. If you've got to hire somebody and pay them to run it and administer it, how much are you saving at the end of the day?"
Citing Asterisk users such as Yahoo!, Google and Sun Microsystems (now Oracle), Sokol disputed the claim, saying, "There's absolutely nothing in the design of Asterisk or in the implementation of that prevents you from building a large system."
Let us know what you think about the story; email: Jessica Scarpati, News Writer