Crimson Project Update

on YouTube ↗32m

" summary

A ground-up rewrite of the ceph-osd targets cutting per‑IO CPU cost by moving from thread-pool handoffs to a user‑space scheduler so a single core handles a PG’s messenger, PG state and object store; claimed per‑IO cost targets come from comparing ~20M cycles/I/O (HDD) 300k (SSD) and ~6k (NVMe). Uses c-star futures-based scheduling, a new c-store object store built for ZNS append zones, supports RBD and snapshots, has Crimson rados tests gating PRs, and is adding multicore PG placement, scrub and tiering.

AI-written summary · about

" chapters (8)

1. 0:00Introduction to Crimson
2. 2:47Crimson project goals
3. 5:14Classic OSD threading vs Crimson
4. 7:19Exploiting ZNS storage
5. 9:28C-store object store design
6. 15:01Reef progress and status
7. 18:08Next steps for Crimson
8. 23:57Q&A on tiering and storage