Yank: Enabling Green Data Centers to Pull the Plug
summary
Yank introduces a transient server abstraction plus a snapshot manager and backup engine implemented in Xen to move dirty memory and disk state to multiplexed backups within a short warning window. Prototype on 20 blades (4 GB) demonstrates one backup server can support 15 transient VMs with half its memory free. With 10–20 second warnings Yank cuts transfer volume up to 70× and cuts client response latency about 20× versus zero‑warning HA, and it restores VMs to stable hosts on warning.
AI-written summary · about
chapters (15)
- 0:00Yank
- 4:22Objectives
- 4:57Transient Server Support
- 6:30Existing Approaches
- 8:18Yank Overview
- 9:51Snapshot Manager
- 11:17Backup Engine
- 12:32Green Data Center
- 13:52eval
- 14:32Results
- 16:39Endtoend experiments
- 18:22Conclusion
- 19:02Questions
- 23:20Cool Ideas
- 23:39Summary
related talks
28mLive-Update: An Update on the Journey to Replace the Hypervisor with Limited Downtime
20mFaster Backups, Less Storage: Unlocking Incremental Backups in KubeVirt
27mXen: This Is Not Your Dad's Hypervisor!
20mFrom Agent-Based to Agentless...
15mHyrax: Fail-in-Place Server Operation in Cloud Platforms
38mYour Own Little HA Cloud at the Edge (or at Home)
16mUnlocking unallocated cloud capacity for long, uninterruptible workloads
18mThe Cloud Myths, Schemes, and Dirty Little Secrets