An ip_conntrack hash bucket analysis

This directory contains some pictures made from ip_conntrack hash bucket occupation statistics. We vary the number of buckets, and we vary the hash function used.

The hash functions: abcd abcdx jenkins2 jenkins2b random
The dataset: 50000

• Compare hash functions with fixed number of buckets.
  • about 8191 buckets
  • about 16383 buckets
  • about 32767 buckets
  • about 65535 buckets
  • about 131071 buckets
  • about 262143 buckets
• Compare bucket sizes with fixed hash function.
  • about 8191 buckets
  • about 16383 buckets
  • about 32767 buckets
  • about 65535 buckets
  • about 131071 buckets
  • about 262143 buckets

The normalized maximal bucket length

The sum of the normalized maximal bucket lengths

The standard deviation (sigma-square):

The sum of the standard deviations (sigma-square):