Allocating Less: Really Thin Rust Cloud Apps
summary
Shows concrete memory wins and tools: heap-profiles with dhat, JSON comparisons that cut peak heap by ~30% using json-rust, and a repo with benchmarks. Recommends flatbuffers (no-deserialize paths) to avoid copies, enum_dispatch to remove per-call Box allocations, and cargo-bloat to shrink binaries. Gives measured costs (Arc clone ≈30–120 ns), suggests bumpalo arenas for scoped allocations, and compares jemalloc vs mimalloc for allocator tradeoffs.
AI-written summary · about
chapters (4)
related talks
54mAllocating Less: Really Thin Rust Cloud Apps
51mAllocating Less: Really Thin Rust Cloud Apps
31mBenchmarking and Optimization of Rust Libraries
04m“Profiling Rust Memory Usage with jemalloc”
40mWhy Rust?
27mImproving Rust Performance Through Profiling and Benchmarking
32mBuilding and Maintaining Rust at Scale
29mWrangling memory leaks where you thought they could never happen