Packeteer's new IntelligenceCenter application and WAN management toolset is just part of a broader strategy that should tie in with industry moves towards the sharing of network performance data, according to Simon Jackson, the WAN specialist's European director of system engineering.

While the Packeteer app will make it easier for customers to optimise application performance by taking in NetFlow data from other network devices, as well as reports from its own PacketShaper, iShaper and iShared WAFS boxes, Jackson said that the company is also opening up its performance data to other analysis engines.

"We have moved away from a proprietary Oracle database to the open-source PostgreSQL database, with a binary input to stream data straight in - it gives us both performance and scalability," he said.

Publishing performance data

"Next we will publish APIs and deliver an interface on top of that. It's something I'm a big proponent of - letting others make use of the data we collect. It is a big step for us."

He added that the first software company to make use of this new openness is network behaviour specialist Mazu Networks, which will mine the Packeteer data as part of its intrusion detection work.

"We are also talking to people about using it for billing, anomaly detection, and so on," he said.

"The big integrators are adept at mining data - all they need is SQL, so the drive is for us to keep to SQL syntax and probably provide stored procedure access to control how queries are built" - the latter to hopefully avoid a badly-built query overloading the system.

As networks become more complex, and as more and more different devices are deployed within them, Jackson said it is vital to make use of the data those devices are already generating - not least because if the instrumentation is already there, it could save the cost and complexity of adding a whole extra layer of monitoring tools.

"The information is there in our systems, you just have to get to it," he said.

It is also an acknowledgement by Packeteer that no matter how good its devices are at reporting on bandwidth usage, they are passive so cannot collect all the flow data available. That's why IntelligenceCenter needs to be able to take NetFlow data as well.

The need for openness and flexibility is being acknowledged elsewhere too, he claimed. For instance, Packeteer is shifting from building its user interfaces in HTML towards Adobe Flash and Flex, as he said they offer better cross-platform compatibility and make it possible to do rapid prototyping.

And in the future, it could enable new Web 2.0-type ways of doing things, such as drill-downs and mash-ups that take in data from all these different sources and present it in new ways, he predicted.

PacketShaper goes Turbo

Jackson said that Packeteer is also adding a new high-end PacketShaper 10000 aimed at service providers and large data centres. Capable of fully forwarding 6Gbit/s of traffic - so three full-duplex 1Gig feeds - the device is built around a network processor chip (although he wouldn't name the chip supplier), and is the first in a new family, imaginatively named Turbo.

The aim is to put Turbo up against the likes of Allot, Cisco and Sandvine with a device capable of shaping and controlling multi-Gigabit traffic flows without dropping packets, a task that requires you to use deep packet inspection to identify flows before you apply controls to them.

Jackson claimed that while rivals such as Expand, Juniper and Riverbed have grabbed the headlines for their WAN acceleration appliances, their technologies will eventually become just another standard part of the network - unlike flow-shaping technologies.

"The industry has been so skewed towards [WAN] compression and acceleration devices that it has forgotten about delivering a deterministic service," he declared. "Eventually, all those performance technologies become commodities, but they don't give you business advantage."

That business advantage comes from application performance, and to improve that you need to measure and correlate network, cache and application data, he added.

"We want to be involved in performance everywhere, from big data centres to the edge and onto mobile devices," he said. "It's all about your combined collection of data, and finding the value in it."