aboutsummaryrefslogtreecommitdiffstats
path: root/research/experiments/013-parallel-syncs/2026-09-27-store-variants-ext4-linux-aarch64.jsonl
Commit message (Collapse)AuthorAgeFilesLines
* Measure what moving saves off the owner would buy (084)eunhhu2026-09-271-0/+21
Nothing a thread can give: with writes saturated the device's one line of syncs is the bottleneck, so zeroing the tail or compacting on another thread only queues that device work behind the commits, and the owner's idle gap is shorter than one journal commit. Built without compaction the store is no faster; without compaction and without tail growth durable posts gain 1.4% and the p99 drops 24% (17.4 to 13.3 ms). New blocks are the tail latency, so the way to it is a store that rewrites its space in place, a ring cleaned forward instead of compacted into new files.