Details
-
Improvement
-
Status: Closed
-
Major
-
Resolution: Duplicate
-
1.3.0
-
None
-
None
Description
It would be great if we could checkpoint operator state asynchronously. That way we would not risk blocking the processing of elements due to a file system write quota, for example.
Being able to asynchronously checkpoint operator state entails that we can obtain a copy of the original state in case of a concurrent write operation.
Attachments
Issue Links
- duplicates
-
FLINK-6048 Asynchronous snapshots for heap-based operator state backends
-
- Closed
-