The HTTP header parser and management code inherited from Squid allocates individual structures (HttpHeaderEntry) for each header being parsed and represented in a HTTP request/reply. This is an inefficient use of CPU and memory.
This project aims to rewrite the HTTP header management code to use a more compact representation in memory.
The specific goals include reducing the amount of calls to the memory management routines and the subsequent drop in CPU utilization. Instead of 1 call to allocate per HTTP header entry, a single call to allocate as many entries as needed.
The changes will be linked to Issue 108 in the bug tracking system.