The IPv6 Interceptor (red) provides significant performance advantage over conventional scheme (yellow) since it do not need to check or fully decode all contents. The green is the performance of the router in no intercept mode.

Fast Stream Interceptors for Internet Filter Appliances

Stream interception is rapidly becoming a common task in Internet appliances. Beginning from cache, proxy, filters, firewalls and gateways, there are now host of new services including content adaptation, content personalization, location-aware data insertion, to security filters -- all are fundamentally stream interception machines requiring some form of intermediate access inside transiting traffic’s content. A significant percent of the delivered Internet traffic is now ‘touched’.
However, if we look into the current protocol design, particularly the protocol packet structures we will see that it has little facility that enables efficient stream intercept.

Most real-life content, particularly high performance networked multimedia transport data carried over a network packet are multi-level hierarchically encapsulated (that means bags within suitcases). For example an H.261 video element may require access to 16 pre-dependent elements before the element of interest can be read.

 

 

WW: Javed I. Khan & Yihua He, Embedded Data Indexing for Fast Stream Interception by Internet Appliances, Proceedings of the IEEE/WIC International Conference on Web Intelligence, WI'03, October 2003, Halifax, Canada, pp.579-583. [KhHe03]

 

WW: Javed I. Khan and Yihua He, A Framework for Active Ubiquitous Application Services,, 6th IASTED International Conference on Internet Multimedia Systems and Applications, IMSA 2002, Kauai, August 2002, pp.145-151. [KhHe02], (GS(1)).

 

WW: Javed I. Khan & Yihua He, Fast Intercept of a Passing Stream for High Performance Filter Appliances, 5^th IEEE International Conf. on High Speed Networking and Multimedia Communications, HSNMC2002, Jeju, Korea, July 2002., pp122-127. [KhHe02]

Selected Publications

More Publications >>

The problem has major implication on the CPU cycle, memory size, and overall performance of any intercepting appliance system’s architecture. Some form of random access will be one of the most important factor for high performance stream data processing appliances.

In this research we are investigating this new but important problem and demonstrate a novel content indexing scheme that can facilitate dynamic index based random access into streams and provide performance boost to intercepting filter like appliances. It is perhaps as salient to appliance’s overall architecture as the design of disk scheduling algorithm or multilevel memory/cache organization is to the conventional machine architecture.

Publications

Distributions

Demo

Technical Reports

All Projects

 

Page last updated February, 2008, Medianet Laboratories.