New allocator idea sounds a lot like the metaslab approach used in ZFS.
Since ZoL 0.7, Raid-Z parity and checksumming (Fletcher4) operations are accelerated using SIMD instruction sets (in case of ThunderX, using aarch64 NEON)
New allocator idea sounds a lot like the metaslab approach used in ZFS.
Since ZoL 0.7, Raid-Z parity and checksumming (Fletcher4) operations are accelerated using SIMD instruction sets (in case of ThunderX, using aarch64 NEON)