Important Information
Flink 1.12.0 has been released.
Features
List of features announced by contributors and committers that are likely to be ready for the feature freeze:
Runtime
Feature Name | Proposed by | Note |
---|---|---|
K8s: Support user-specified pod templates | Canbin Zheng | |
K8s: Support built-in init-Container | Canbin Zheng | |
Support accessing secured services via K8s secrets | Canbin Zheng | |
K8s: Support configuring replica of JobManager Deployment in ZooKeeper HA | Canbin Zheng | |
K8s: Support to configure limit for CPU requirement | Canbin Zheng | |
Andrey Zagrebin, Zhu Zhu, Till Rohrmann | ||
Reactive-scaling mode | Till Rohrmann | |
FLIP-138: Declarative Resource management | Chesnay Schepler, Till Rohrmann | |
Chesnay Schepler, Till Rohrmann | ||
Robert Metzger, Till Rohrmann | ||
Robert Metzger, Till Rohrmann | ||
Yangze Guo, Xintong Song, Till Rohrmann | ||
FLIP-76: Unaligned Checkpoints (improvements and follow ups) | Arvid Heise / Zhijiang Wang / Piotr Nowojski | |
Piotr Nowojski | ||
Sub-second end-to-end exactly-once latencies | Piotr Nowojski / Roman Khachatryan | FLIP/Jira issue pending |
FLIP-135 Approximate Task-Local Recovery | Yuan Mei | Milestone One |
FLIP-141: Intra-Slot Managed Memory Sharing | Xintong Song | |
Xintong Song | ||
Runtime: various resource manager improvements. | Xintong Song, Canbin Zheng, Till Rohrmann | |
Yangze Guo | ||
k8s: Support native k8s for sql client, python client and scala shell | Yang Wang, Canbin Zheng | |
Yang Wang, Shengyang Sha | ||
Zhijiang Wang | ||
Improve the performance of batch shuffle | Stephan Ewen, Zhijiang Wang, Yingjie Cao |
Core APIs (DataStream/DataSet)
Feature Name | Proposed by | Note |
---|---|---|
FLIP-131: Consolidate the user-facing Dataflow SDKs/APIs (and deprecate the DataSet API) | Aljoscha | |
Dawid | ||
FLIP-134: Batch execution for the DataStream API | Klou / Dawid / Aljoscha | FLIP/Jira issue pending |
FLIP-140: Introduce batch-style execution for bounded keyed streams | Dawid |
Connector APIs
Feature Name | Proposed by | Note |
---|---|---|
Unified Sink Interface | Guowei / Yun | FLIP/Jira issue pending |
Extend StreamingFileSink to handle the “small file problem” | Jingsong / Yun Gao | FLIP/Jira issue pending |
Becket Qin | ||
FLIP-33: Standardize Connector Metrics | Becket Qin | |
Hybrid Source to support switch over between sources in the same job. | Xiaofeng Jiang | |
Connector Testing Framework | Qingsheng Ren | |
Tzu-Li (Gordon) Tai, Danny Cranmer, Roland | ||
Waldemar Hummer, Alexander Alexandrov, Danny Cranmer |
Table API/SQL
Feature Name | Proposed by | Note |
---|---|---|
Timo Walther | ||
Table/SQL: FLIP-129: Refactor Descriptor API | Jark Wu | |
Table/SQL: FLIP-132: Temporal Table DDL and Temporal Table Join | Leonard Xu | |
Table/SQL: Compiling multi-input operator to reduce unnecessary shuffle | Godfrey He | |
Table/SQL: FLIP-145: Support SQL windowing table-valued function | Jark & Danny | |
Table/SQL: Hive query compatibility | Rui Li | FLIP/Jira issue pending |
Table/SQL: CREATE FUNCTION USING JAR | Jark Wu | FLIP/Jira issue pending |
Table/SQL: FLIP-136: Improve interoperability between DataStream and Table API | Timo Walther | |
Table/SQL: FLIP-107: Handling of metadata in SQL connectors | Timo Walther | |
FLIP-36: Support Interactive Programming in Flink | Xuannan Su |
PyFlink
Feature Name | Proposed by | Note |
---|---|---|
PyFlink: Improve the usabilities for PyFlink | Jincheng Sun | FLIP-133 |
PyFlink: PyFlink Table API DSL | Jincheng Sun | |
Jincheng Sun | FLIP-130 | |
PyFlink: Support Pandas UDAF in batch GroupBy aggregation | Jincheng Sun | FLIP-137 |
PyFlink: PyFlink Table API UDAF | Jincheng Sun | FLIP-139: General Python User-Defined Aggregate Function Support on Table API |
PyFlink: Support running pyflink jobs on kubernetes | Jincheng Sun |