The WAN optimization market is a crowded space these days, so the savvier newcomers target some piece of the action and hammer that piece into a market presence. That's the path followed by Infineta Systems, a four-year-old Santa Clara, Calif.-based startup that's trained its guns on WAN op demands for moving latency-sensitive data and applications at high speed between data centers -- something that's increasingly more important in the era of big data.
Infineta's flagship product is called the Data Mobility Switch (DMS), which starts at about $80,000. It's a hardware-based appliance designed to improve WAN performance and boost capacity by as much as 10X, making it an ideal fit for the movement of data between servers and bandwidth-intensive functions such as live migration and remote backup.
The capacity boost is accomplished through a series of compression and deduplication algorithms that assess network traffic and make it possible for application throughput to exceed a WAN link's stated bandwidth. Applying the DMS' firepower to a 1-Gbps WAN, for example, turns it into a virtual 5-Gbps link, or a 622-Mbps OC12 link into a 3-Gbps link.
The custom hardware is a different approach than that of bigger WAN optimization players using x86-based platforms, but it's necessary to get the level of performance Infineta promises, Haseeb Budhani, Infineta's chief product officer, said in a recent interview with CRN.
"The merchant silicon approach we've taken lends itself nicely to the implementations we're doing," Budhani explained. "WAN optimization, if it's not very easy to deploy, becomes a very difficult sale. You want to come in, prove a point, and walk away. So we need to prove our point."
Infineta also offers a pay-as-you-go WAN optimization service, Data Migration Accelerator, built around the DMS. And in June, it launched FlowTune, a QoS feature set for the DMS that unknots network congestion by dynamically managing application transmission rates.
That's also a big difference from how other vendors address the congestion problem, Budhani said, because many of their solutions require waiting for packet drops to unclog networks, or forcing customers to manage WAN traffic using other means, such as point-to-point circuits, or Layer 2 QoS techniques such as weighted random early detection (WRED) that can stave off traffic surges but also harm performance of other applications.
"Packet loss is a symptom of congestion," Budhani said. "So what if we set up a system where there is no loss and I could prevent that congestion still?"
WAN optimization, which research firm IDC has pegged as a roughly $40 billion market by the end of 2012, is becoming increasingly more competitive as a channel discipline and is a magnet for startups.
As of this year, more than half of large enterprises in North America have some kind of WAN optimization solution in place, according to 451 Research's TheInfoPro, so the major players, such as Riverbed, have sought to diversify into other areas such as application delivery and Web content delivery, while buzzed-about startups such as Aryaka are attacking the problem with cloud-delivered solutions.
Relative to Riverbed, Cisco, Blue Coat and the space's other incumbent players, Infineta is still a newcomer. The company was founded in 2008, came out of stealth mode in May 2010, and has since raised about $30 million in venture capital funding.
Infineta sold primarily direct to start, but a channel presence is now finally emerging, and Budhani said Infineta's ultimate goal is to fulfill 100 percent through the channel. This summer, Infineta joined NetApp's Alliance Partner Program, and in July, it revealed two key reseller relationships, one with Nissho Electronics of Japan and the other with FusionStorm, the San Francisco-based solution provider and infrastructure integrator.
"Why would they work with Infineta? The answer is pretty clear," Budhani said. "Riverbed has a big hold on the branch side and the lower midtier of the market. But when it comes to high-speed needs, with a gigabit or higher, their solution just doesn't pan out. We think a lot of their partners could look at it is as saying, 'If I'm selling to one big company, I can sell Riverbed for 'x' and Infineta for 'y'."
NEXT: FusionStorm's Success With Infineta