IPv6 packet classification based on flow label, source and destination addresses

Packet classification is important for supporting multi-field packet routing, quality of service (QoS), real-time data streaming applications and firewalls. Existing schemes in IPv4 uses the 5-tuple approach, giving much overhead in the dataplane processing. With the introduction of IPv6 flow label...

Full description

Bibliographic Details
Main Authors: Poh, Eric C. K., Ewe, Hong Tat
Format: Book Section
Language:English
Published: IEEE Xplore 2005
Subjects:
Online Access:http://shdl.mmu.edu.my/2393/
http://shdl.mmu.edu.my/2393/1/IPv6%20packet%20classification%20based%20on%20flow%20label%2C%20source%20and%20destination%20addresses.pdf
Description
Summary:Packet classification is important for supporting multi-field packet routing, quality of service (QoS), real-time data streaming applications and firewalls. Existing schemes in IPv4 uses the 5-tuple approach, giving much overhead in the dataplane processing. With the introduction of IPv6 flow label field which facilitates labeling of packets belonging to the same flow; classification can now be done using the 3-tuple approach. In this paper, we propose 3-tuple packet classification algorithm using hierarchical-trie data structure that has low storage requirement and optimum search speed. Packet filters are evaluated based on hierarchical approach, starting from flow label followed by source and destination addresses. A hashing model with collision resolution for addresses that minimizes memory consumption is also being described. The time complexity of the lookup process is bounded by the length of header fields and is independent of the trie size.