Post-Copy steps

  1. suspends the migrating VM at the source node
  2. copies minimal processor state to the target node
  3. resumes the virtual machine
  4. begins fetching memory pages over the network from the source.
    The manner in which pages are fetched gives rise to different variants

trade-off

  • For VMs with read-intensive workloads, pre-copy would be the better approach
  • For large-memory or write-intensive workloads, post-copy would better suited.

Pre-copy是为了能够尽量地减少downtime,并且不影响服务的运作;而post-copy则是为了减少total migration time且让程序能够持续的执行。

资料