Methods for performing packet classification
AssigneeIntel Corporation
InventorsHarsha Narayan, Alok Kumar
Life of this patent
Estimating the term…
Loading patent…
AssigneeIntel Corporation
InventorsHarsha Narayan, Alok Kumar
Estimating the term…
Methods for performing packet classification. In one embodiment, packets are classified using a rule bit vector optimization scheme, wherein original rule bit vectors in recursive flow classification (RFC) chunks are optimized by removing useless bits that have no effect on the ultimate rule identified by an associated RFC lookup process. The unique optimized rule bit vectors for associated chunks are then cross-producted to produce an optimized downstream chunk. In another embodiment, a rule database splitting scheme is employed. Under this technique, split criteria is defined to split a rule database, such as splitting based on a particular field value or range. A respective set of downstream chunks is then generated for each partition, beginning with the chunks in a split phase. The applicable rule bit vectors for the chunks associated with a common group and partition are identified, and then unique applicable rule bit vectors for those chunks are cross-producted to produce downstream chunks.
A method, comprising:
A method comprising:
A machine readable medium, to store instructions that if executed perform operations comprising:
A machine readable medium, to store instructions that if executed perform operations comprising:
The machine readable medium of claim 21, wherein execution of the instructions performs the further operation of:
The machine readable medium of claim 21, wherein execution of the instructions perform further operations comprising:
A network line card, comprising:
Dependencies are inferred from the claim text.