Apache Beam 2.9.0 发布了。Apache Beam 是 Google 在2016年2月份贡献给 Apache 基金会的项目,主要目标是统一批处理和流处理的编程范式,为无限、乱序、web-scale 的数据集处理提供简单灵活,功能丰富以及表达能力十分强大的 SDK。Apache Beam 项目重点在于数据处理的编程范式和接口定义,并不涉及具体执行引擎的实现,Apache Beam 希望基于 Beam 开发的数据处理程序可以执行在任意的分布式计算引擎上。
下载地址:https://beam.apache.org/get-started/downloads/
此版本包括以下主要新特性和改进:
依赖升级
Update google-api-client libraries to 1.27.0.
Update byte-buddy to 1.9.3
Update Flink Runner to 1.5.5
Upgrade google-apitools to 0.5.24
I/Os
I/O connector for RabbitMQ.
Update SpannerIO to support unbounded writes.
Add PFADD method to RedisIO.
其他修复
Dataflow runner was updated to not use Conscrypt as the default security provider.
Support set/delete of timers by ID in Flink runner.
Improvements to stabilize integration tests.
Updates Spark runner to show Beam metrics in web UI
Vendor gRPC and Protobuf separately from beam-model-* Java packages
Avoid reshuffle for zero and one element creates