VMFT - Primary-backup replication
Overview
- virtual machines layers
- replication in different layers
- e.g., replicating GFS master
Approaches
- pause, replicate states, continue
- easy to implement
- state machine, replicate inputs
- efficient in performance
Non-determinism
- clock
- record
- multi-core (different from multi-thread)
- not supported in vmft
- IO race
- non-blocking
- DMA
- exactly once over network? (TCP, UDP)
- disk read
Network partition and “split-brain”
- failure detector
- atomic compare and swap
- can lease solve the problem?