Wednesday 29th August, 2012
11:40am to 12:25pm
There is a strong desire to consolidate the benefits of the three Linux slab allocators, SLAB, SLUB, and SLOB, into a single allocator. Choosing amongst these allocators is difficult when systems run a variety of workloads. SLAM is the next generation slab allocator for Linux and is a drop-in replacement for the current allocators without sacrificing performance or memory footprint. SLAM enables caches to utilize different slowpaths based on their use so they operate most efficiently with an optimized fastpath.
This talk appeals to a technical audience because virtually every subsystem within the kernel relies upon slab. Developers need to be aware of the new interfaces available to optimize their memory allocations based on how their subsystem uses kernel memory. They also have a desire to reduce the variance in performance as the result of which allocator is configured.
senior software engineer for Google, Inc
Sign in to add slides, notes or videos to this session