Github user sihuazhou commented on the issue:
https://github.com/apache/flink/pull/6301
Hi @jrthe42, In general the checkpoint include two part of works.
- part1: take a snapshot of the state.
- part2: transfer the snapshot to the checkpoint destination(e.g. DFS)
The part1 need to be sync, and the part2 can be async, if I'm not wrong.
---
