To Kill, or to Checkpoint - That is the Question
Cloud native containerized applications heavily rely on resiliency. When a container dies, it is restarted. When a node is under evacuation, containers it hosts are killed and startedelsewhere. But th …
| Talk Title | To Kill, or to Checkpoint - That is the Question |
| Speakers | Mike Rapoport (Researcher, IBM), Adrian Reber (Principal Software Engineer, Red Hat) |
| Conference | Open Source Summit + ELC Europe |
| Conf Tag | |
| Location | Edinburgh, UK |
| Date | Oct 21-25, 2018 |
| URL | Talk Page |
| Slides | Talk Slides |
| Video | |
Cloud native containerized applications heavily rely on resiliency. When a container dies, it is restarted. When a node is under evacuation, containers it hosts are killed and startedelsewhere. But there are occasions when checkpointing and restoring a container would be more effective. Long-running jobs, stateful applications, function runtimes for serverless computing, all could benefit from checkpoint-restore and container migration. And for some applications, the absence of these capabilities could be a show stopper for moving from virtual machines to containers.With CRIU, the industry standard container migration tool, it is possible to migrate containers and reboot hosts without losing network connections, results of long-running calculations, or the state ofin-memory databases.In this talk, we present the container migration technology and its applicability to various use cases.
