US 7,466,687 B2
Packet classification using modified range labels
Claude Basso, Raleigh, N.C. (US); Jean Louis Calvignac, Cary, N.C. (US); Gordon Taylor Davis, Chapel Hill, N.C. (US); and Clark Debs Jeffries, Durham, N.C. (US)
Assigned to International Business Machines Corporation, Armonk, N.Y. (US)
Filed on Apr. 28, 2003, as Appl. No. 10/425,097.
Prior Publication US 2004/0213275 A1, Oct. 28, 2004
Int. Cl. H04L 12/28 (2006.01)
U.S. Cl. 370—351  [370/389; 370/395.54; 370/411] 7 Claims
OG exemplary drawing
 
1. A method of processing parameter ranges for each parameter field in a packet classification rule, said method comprising:
(a) associating at least one rule with at least one dimension, said dimension being a parameter in a network packet, said at least one rule being an instruction to a router for processing said network packet;
(b) sorting said at least one rule according to a dimension associated with each set at least one rule;
(c) grouping said at least one rule with a second at least one rule into a cluster of rules with overlapping ranges, overlap of said ranges being determined by said at least one dimension;
(d) upon determining that said at least one rule overlaps said second at least one rule, expanding said cluster of overlapping rules into multiple unique segments that identify unique range intersections;
(e) expanding the number of said multiple unique segments such that each rule has a number of segments that is a power of two;
(f) offsetting each said cluster to an optimum alignment position;
(g) converting the numbering of said range segments from binary to Gray code to create Gray coded range segments;
(i) identifying a unique rule corresponding to a group of said Gray coded range segments; and
(j) utilizing the unique rule to define and instruct how a particular packet is to be routed.