Lightning Talk: Containers Live Migration: What’s There, What’s Missing, What’s Next?
summary
Live container migration exists in pieces: CRIU (used by Docker/Podman and even Google Borg) and DMTCP can checkpoint process state, DockerCon 2015 demo moved Quake3 but relied on SCP and hacks, and Kubernetes adds a checkpoint API (1.25 alpha, later beta) that only saves tarred checkpoints on the node — there is no seamless restore/transfer across hosts yet. Missing pieces include CNI-aware restore, discless/lazy page transfer, encrypted checkpoints, and operator-based pod-migration tooling that moves and restores checkpoints end-to-end.
AI-written summary · about
chapters (11)
- 0:00Containers live migration intro
- 1:51Challenges in live migration
- 3:59Use cases for migration
- 5:48History of demos
- 6:31Tools for process state
- 7:59CRIU as checkpointing standard
- 9:03Kubernetes limitations
- 9:47Missing restore and migration
- 10:31Future requirements
- 10:53Next steps and improvements
- 12:19Summary and references
related talks
39mContainer Live Migration
43mContainer Live Migration in Kubernetes Production Environment
26mKubernetes and Checkpoint Restore
24mOptimizing K8s and unlock blockers with Pod Live Migration
11mLightning Talk: How Jenga & Tetris Can Help Run More Containers with Lesser Costs & Downtime
39mKubernetes and Checkpoint/Restore
34mLetting Your Containers Dream of Electric Sheep
43mPanel Discussion: Containers and Networking: A Symbiotic Relationship