davidradl commented on code in PR #27489:
URL: https://github.com/apache/flink/pull/27489#discussion_r2740769229
##########
docs/content/docs/ops/state/disaggregated_state.md:
##########
@@ -204,4 +204,25 @@ Default is true. Setting this to false will raise the CPU
usage.
- `state.backend.forst.executor.inline-coordinator`: Whether to let task
thread be the coordinator thread.
Default is true. Setting this to false will raise the CPU usage.
+{{< hint info >}}
+`ForStStateBackend` utilizes [ForSt](https://github.com/ververica/ForSt/) as
its underlying database core.
+The current iteration of ForSt is forked from
[frocksdb](https://github.com/ververica/frocksdb),
+architected as an embedded database core specifically for Flink local state
management.
+
+While transitioning the db toward a disaggregated architecture, we encountered
significant
+architectural and engineering constraints within the existing framework.
+To address these challenges, we are now actively developing a next-generation,
Review Comment:
This is curious as the code referred to is in a non Flink repo, so the we
really means Ververica not Flink. Is there an intension to move Forst to a
Flink repo, so this becomes business as usual.
In the spirit of openness I think we should document or point the docs to
the "significant
architectural and engineering constraints within the existing framework.".
So users can take an informed view as whether to adopt this at this time.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
