Honestly, I am surprised by the great response to OpenDPI. Two weeks after releasing version 1.0 we have counted 1000 downloads. This is much more than we expected for this very special piece of software.
Making the own intellectual property public is an interesting experience. Of course we had a long internal discussion about that irreversible step beforehand. What do we give away and what will we get back? What will other people say about our software? But all responses showed that it was the right step.
Last week, I participated in a workshop session called “Deep Packet Inspection: Technology, Promise & Controversy. What You Need to Know.” at the Broadband World Forum Europe 2009 in Paris. The panel discussion with participants from DPI vendors, Internet service providers and net neutrality activists was meant to “foster an open, balanced and rigorous discussion of DPI’s capabilities, benefits, limits and concerns”.
Many Internet users worry about net neutrality and online privacy these days. And they see deep packet inspection (DPI) as their arch enemy. Common claims are that DPI reads the content of all packets and then decides based on the nature of the content whether to forward, slow down or drop that packet. This is a misconception based on poor understanding how today’s DPI for Internet traffic management works.
DPI is only the core technology of traffic management systems that is used for application classification. This is done by scanning the first few packets — including their payload — of each network flow for certain patterns out of a predefined list. DPI does not ‘read’ or even ‘understand’ any communication content. As soon as the communicating application of a network flow is identified, DPI stops for that flow. Then, bandwidth management rules determine how this flow is treated. Possible rules are:
Honestly, I am surprised by the great response to OpenDPI. Two weeks after releasing version 1.0 we have counted 