apache/beam

[Feature Request]: Cache the BigQuery table definition instead of calling tables.get() from every worker

Open

#34,076 创建于 2025年2月26日

在 GitHub 查看
 (9 评论) (0 反应) (2 负责人)Java (7,313 star) (4,097 fork)batch import
P2good first issuenew featurepythonstale

描述

What would you like to happen?

When calling BigQuery's tables.get method, today, we are calling it from every Dataflow worker. This is resulting in customers hitting Quota issues. Please can we cache the Table definition and re-use for all workers instead of calling the method from every worker.

https://cloud.google.com/bigquery/docs/reference/rest/v2/tables/get

https://github.com/apache/beam/blob/9a1e7bf4cd399177e5de7fe28bad741cade452a5/sdks/python/apache_beam/io/gcp/bigquery.py#L902-L903

https://github.com/apache/beam/blob/9a1e7bf4cd399177e5de7fe28bad741cade452a5/sdks/python/apache_beam/io/gcp/bigquery_tools.py#L565

Issue Priority

Priority: 2 (default / most feature requests should be filed as P2)

Issue Components

  • Component: Python SDK
  • Component: Java SDK
  • Component: Go SDK
  • Component: Typescript SDK
  • Component: IO connector
  • Component: Beam YAML
  • Component: Beam examples
  • Component: Beam playground
  • Component: Beam katas
  • Component: Website
  • Component: Infrastructure
  • Component: Spark Runner
  • Component: Flink Runner
  • Component: Samza Runner
  • Component: Twister2 Runner
  • Component: Hazelcast Jet Runner
  • Component: Google Cloud Dataflow Runner

贡献者指南

技术栈
pythongcp
领域
backenddatacloud
议题类型
feature
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
4
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
1-2 days
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
blocked
清晰度议题是否清楚说明期望改动、验收标准和下一步。
clear
前置要求
PythonBigQuery APICachingApache Beam SDK
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
35
研究方向
Investigate the existing calls to tables.get() in the Python SDK, specifically in bigquery.py (line 902-903) and bigquery tools.py (line 565) as linked in the issue. Determine a caching strategy that avoids calling the API from every worker. Consider using a shared cache across workers or a lazy initialization pattern. Review the 9 comments in the issue for any proposed solutions or maintainer feedback.