Hi all, Dong(cc'ed) and I are opening this thread to discuss our proposal to support dynamically triggering checkpoints from operators, which has been documented in FLIP-309 <https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=255069517>.
With the help of the ability proposed in this FLIP, users could improve the performance of their Flink job in cases like when the job needs to process both historical batch data and real-time streaming data, by adjusting the checkpoint triggerings in different phases of a HybridSource or CDC source. This proposal would be a fundamental component in the effort to further unify Flink's batch and stream processing ability. Please feel free to reply to this email thread and share with us your opinions. Best regards. Dong and Yunfeng