Another interesting implementation that is written by an apache http server, APR, and subversion developer is pocore: https://github.com/gstein/pocore/blob/mas

Another interesting implementation that is written by an apache http server, APR, and subversion developer is pocore: https://github.com/gstein/pocore/blob/master/src/memory.c https://github.com/gstein/pocore/blob/master/src/memory.c https://github.com/gstein/pocore/blob/wiki/MemoryManagement.md https://github.com/gstein/pocore/blob/wiki/MemoryManagement.... It's design is heavily influenced by real world experiences -- as this post even changed its benchmarks to reflect that in an HTTP server, a request uses maybe 100 small chunks per request, you can make different optimization choices.