x
Optical/IP

Cisco Lags DPI Rivals

A recent independent lab test of Deep Packet Inspection (DPI) appliances provided a mix of results for Cisco Systems Inc. (Nasdaq: CSCO) -- its technology passed the tests, but its appliance could only only take in half the traffic of its competition.

There's no deep technology mystery behind Cisco's shortfall -- it's a matter of port count. Cisco's gear doesn't have as many 10-Gbit/s ports as the boxes offered up by the other companies involved, ipoque GmbH and Procera Networks .

The three vendors were included in an Internet Evolution test of peer-to-peer (P2P) traffic detection, run by the European Advanced Networking Test Center AG (EANTC) .

All three of the tested appliances -- the Cisco SCE-8000, the ipoque PRX-10G Traffic Manager, and the Procera PacketLogic 10014 -- performed well when it came to the esoteric tests that engineers have to run. But in the less esoteric "How many ports can I connect?" test, Cisco flopped: It could only hook up to four 10-Gbit/s Ethernet load modules on an Ixia (Nasdaq: XXIA) chassis, while the others could do eight. (See Cisco Plays Catchup in DPI Test.)

(Each vendor paid to participate in the test and was given the option to drop out after seeing the results and EANTC's analysis. No vendor was shown the other vendors' results or even told which other boxes were included.)

Due to practical constraints, EANTC could run just 6.25 Gbit/s worth of throughput through each pair of 10-Gbit/s interfaces. So the ipoque and Procera boxes got tested under some 25 Gbit/s cases (four pairs of interfaces), while Cisco's tests maxed out at half as much.

That's not to say Cisco's P2P filtering isn't ready for prime time. On the contrary, one conclusion of the test was that all three companies' gear is suitable for deployment in carrier networks in multi-gigabit-per-second cases (some more "multi" than others, admittedly).

Judging from his report, EANTC managing director Carsten Rossenhövel seems quite impressed with improvements in DPI hardware in general, and with good reason. EANTC ran a similar test in 2008 but could only do 1-Gbit/s speeds. Even then, few vendors agreed to participate in the test, though it was free. (See Peer-to-Peer Filters: Not Worth the Hype?)

P2P detection has come far quickly, but the same can't be said for content filtering -- the technology that could monitor a stream and pick out copyrighted traffic, for instance. Companies such as Audible Magic and SafeMedia Corp. declined to participate in EANTC's tests. Word has it the products don't work in large-scale environments, Rossenhövel writes.

The full report of EANTC's "P2P Taste Test" can be found here.

— Craig Matsumoto, West Coast Editor, Light Reading

Be the first to post a comment regarding this story.
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE