File sharing beyond the speed of light

It sounds like a script from the late 1990s. An unprofitable San Francisco company, with a two-year-old product named after a fish, reports revenues growing at 250% and a blue-chip roster that includes LG Electronics, Liz Claiborne and the Royal Navy. It puts up 9 million shares at its initial public offering (IPO); investors place orders for 250 million and hand it $87 million – which, it says, it neither needs nor intends to spend. The stock lists at $9.75 and soars to $35 within weeks, generating a market cap of $2 billion.

The company is Riverbed Technology, the fish is a Steelhead trout and the profile dates from September 2006, not 1999.

In an era when high-tech listings are infrequent – and blockbuster IPOs even rarer – there is a clear perception that Riverbed has something very special.

Jerry Kennelly, its co-founder and CEO, certainly thinks so: “We are addressing the biggest untapped market in technology today.”

That opportunity stems from the fact that many organisations want their Internet-based wide area networks (WAN) to perform just like their local area networks – whether they are sending a word document locally or transferring a 60MB file globally.

“The secret problem in network connectivity is that the speed of light is not fast enough,” says Kennelly. Notwithstanding the fact that data is zapped through fibre-optic networks, current Internet protocols slow the delivery rate to 100 milliseconds per packet – and that is just not fast enough for large files. Riverbed’s patented Steelhead technology sits at each end of a WAN connection, massively increasing the transfer rate.

For dispersed teams – such as engineers sharing designs, architects passing around drawings or legal teams jointly working on huge files of case notes – such WAN acceleration is a boon.

The technology is at its most powerful when organisations want to move big files around distributed offices and come up against high-latency and relatively low bandwidth, says Kennelly. But, the functionality is intrinsically horizontal, he insists: five years from now there will be a Steelhead in every branch office, in many cases eliminating the need for local servers.

LG Electronics has so far invested in $3 million worth of Steelhead appliances, and reports a payback of $557,000 per month. US law firm Allen Matkins shelled out $69,000 for Steelhead plus $16,000 a year for related expenses, but estimates a saving of $980,000 in productivity gains annually by slashing the time attorneys were previously kept waiting for files.

“Litigation case files, which took more than 22 minutes to download, can now be downloaded in two and a half minutes,” says Frank Gillman, Allen Matkins’ director of technology. Since Steelhead shipped in May 2004, Riverbed has sold 14,000 units to 1,300 customers. Prices start at $3,500 for a five-user set-up.

But Riverbed is not alone in spotting this untapped opportunity. Cisco (through its acquisition of FineGround Networks) and Juniper (through the buy-out of Peribit ) are two stalking giants. But others players include Packeteer, Expand Networks, F5 Networks and Citrix (via Orbital Data).

Kennelly is sanguine about the possibility that a powerhouse such as Cisco could try to railroad the market by incorporating such acceleration functionality into its routers. “You need processor power for acceleration, so you can’t simply embed the technology in a router,” he suggests.

With revenues in the region of $90 million for 2006 and “within spitting distance of profitability”, Riverbed is aiming high. It is planning to release a software version of its core product in mid-2007. If Kennelly is right, that will put Steelhead on laptops everywhere.

Pete Swabey

Pete Swabey

Pete was Editor of Information Age and head of technology research for Vitesse Media plc from 2005 to 2013, before moving on to be Senior Editor and then Editorial Director at The Economist Intelligence...

Related Topics