Ceph QoS: How to support QoS in distributed storage system
summary
Ceph gets concrete QoS: SK Telecom implements and upstreams queueing features to stop background rebalancing and recovery from starving tenants and breaking SLAs. The code adds five queue types including client priority queues and an in-crowd per-pool priority queue under development, plus a scheduler inspired by fair queuing that supports reservation and limit-based control and proportional shares. They deploy it in a private cloud for VDI and multimedia on SSDs and report it controls noisy neighbors without disabling self-healing.
AI-written summary · about
chapters (5)
- 0:00Introduction and motivation
- 3:01Challenges in shared storage
- 5:58QoS solutions in Ceph
- 8:37DMClock design and features
- 10:43Q&A and future plans
related talks
44mCeph QoS Refinements for Background Operations Using MClock
49mIntro to Ceph, the Distributed Storage System
02h 59mCeph Storage: a Caffeinated Primer
35mMaking Ceph Fast in the Face of Failure
34mModernizing Ceph Deployments at CERN: CephFS and Object Storage Across Data Centres
29mCeph Operations at Scale
25mSupporting 3 Availability Zones Stretch Cluster
38mEfficient Ceph Performance Troubleshooting in Production Using eBPF