The Cupertino, California-based company is best known for its optimization/acceleration technology, a sector in which it was a pioneer. However, the inspection capability has always been an inherent part of what Packeteer does. Packeteer CEO Dave Cote considers it to be a differentiator vis-a-vis its competitors in optimization/acceleration who he said offer brute force acceleration of whatever is on the pipe. Without visibility you may fill pipe with FTP or iTunes and squeeze out your SAP traffic, he said.
Cote said Packeteer still derives 10% of its revenue from monitoring where an enterprise deploys one of its boxes to gain better knowledge of what is running on its network and how applications are performing. [Now] enterprises are talking to us about using our multi-Gb visibility and potentially even using us on their LAN, he said. Packeteer is now developing the capability to apply its inspection to higher bandwidths. Enterprise networks are now going GbE to the desktop, while their backbones are increasingly 10Gb unsaturated, said Cote.
The company might have to break with its tradition of relying entirely on Intel server silicon to power its boxes, and Cote acknowledged that it is looking at some hardware assist. You’ll need increased performance, which means more horsepower and, potentially, some specialized horsepower for offloading things like initial, port-based packet inspection and the Layer-2/3 stuff, he said.
He said the company does not expect to use any of the companies providing multi-Gb inspection capabilities in the market, such as Bivio or Endace. Our chief engineer was formerly at CloudShield [another high-bandwidth DPI developer, though not one that pursues OEM] and we’ve looked at those technologies, but we don’t think they’re right for what we want to achieve, he said.
Cote was also keen to differentiate what Packeteer calls DPI from how its competitors use the terms. This is not just looking at URLs or string matching, he said. It’s true L7 DPI, which means things like identifying Oracle by the database name to prioritize financial data traffic on days near the end of the month, or prioritizing Citrix print traffic even though it’s not delay-sensitive, in order not to hold up Citrix file stuff.
The development of a multi-Gb inspection device will also take Packeteer into a different type of competition from its traditional optimization/acceleration technology where it goes up against products from companies such as Cisco, Riverbed, and Juniper. It will take us more into the service provider side where we’ll be competing a little bit with what Allot does there, as well as with Cisco’s Service Control Engine technology from its P-Cube acquisition, he said.
Our View
It is interesting that Packeteer is feeling the pull toward DPI for visibility rather than just to provide the raw material for prioritization, optimization, and so on. Bivio, Endace, and OpenShield are already on this tack, though CloudShield currently integrates vertically with its own IDS/IPS capabilities, and Endace plans to offer an IPS module later this year. Other players in this space include Force10 with the technology it acquired when it bought Metanetworks, and of course Cisco offers it with SCE. The next question is: which silicon provider will it select for the offload engine?
