The Big Idea at Acme Packet: Smoothing the Way for Voice and Video on the Internet
You’re an unfunded early-stage startup and your sole venture backer sends you a check for millions of dollars. Unfortunately, it’s made out to the wrong company name, so you can’t deposit it. What do you do? You change your name, of course.
That’s how Primary Networks became Acme Packet. The Burlington, MA, maker of Internet switching equipment and software was founded in 2000 and went public in 2006—today you’ll find it on the NASDAQ exchange under the ticker symbol APKT. But when it got its first capital infusion from Menlo Ventures in Menlo Park, CA, says co-founder and CEO Andy Ory, it was still so new that it hadn’t even printed business cards.
“We’d been using the name Acme Packet in all our presentations, because we thought it was funny, but [Menlo] actually took the name seriously,” says Ory. “They sent us a $12 million check made out to Acme Packet. We changed our name that day.”
In a way, the company’s whole history is about adapting to change. The Acme part of the name may or may not refer to the fictional company that supplied Wile E. Coyote with the rockets, anvils, and other gear he used to pursue the Road Runner. (Ory won’t say, perhaps out of concern for treading on Warner Bros. trademarks.) But the “packet” part definitely refers to the packets into which all data is divided before it can cross the Internet—and Ory and his co-founder Patrick MeLampy saw early on that if the Internet were ever to become a medium for real-time communications such as voice calls and video conferencing, the way these packets travel would have to be rethought. Their big idea: If Acme could come up with faster, more reliable way to shepherd high-priority packets through the existing Internet, telecom providers would beat a path to their door.
[Editor’s Note: Every startup has a “big idea” that it thinks will catapult it to success. With this story, we continue an occasional column highlighting the big ideas—and the resulting challenges—at companies in Xconomy’s home cities.]
Traditionally, Ory explains, the thousands of Internet Protocol (IP) packets that may comprise a file such as an e-mail message, a photo, or a spreadsheet are dispersed and sent across the net along unpredictable paths before being reassembled at the destination address. Packets often get lost along the way, meaning replacements have to be sent. Engineers call this approach “best-effort” networking.
Meeting with me at the company’s headquarters last week, Ory held up a piece of paper. “If I dip this document in liquid nitrogen and break it into 10,000 packets and put them into the network like little pachinko balls, it doesn’t matter what order they arrive in—eventually the sequence gets completed, and you’ve got your e-mail. That’s great, and it usually takes less than 500 milliseconds. But now imagine you and I are doing a voice-and-video-over-IP call, and 24 times every second, a visual of me is being dipped in liquid nitrogen and shipped as packets. They can’t arrive out of order, or with more than 200 milliseconds of delay.”
Best-effort networking just isn’t good enough to handle unidirectional communication with that kind of alacrity, let alone bidirectional communications, Ory explains. So if Voice-over-IP (VoIP) telephony were every to be a reality—and back in 2000, that was still an open question—something new would be needed.
Ory and MeLampy came up with the idea of installing special boxes at the boundaries between the hundreds of privately owned networks that together make up the Internet. These boxes would act as way stations. The packets in a voice or video stream—what Ory calls “signaled media”—could still travel dispersed paths through the cores of the local networks. But at the edges, the way stations would reconstitute them, like regiments of infantry forming up for battle, before handing them off to the next network.
“What emerged was an infrastructure where … Next Page »