0 views collected around this technical thread.
This article details the FlinkSQL performance enhancements implemented by the streaming team, covering view reuse, redundant shuffle removal, multiple‑input operator redesign, long sliding‑window optimizations, and native JSON format improvements, which together deliver up to 60% CPU savings and massive core‑hour reductions.
This article explains how to derive field‑level data lineage for FlinkSQL by leveraging Apache Calcite, covering the Calcite framework, FlinkSQL execution stages, the three‑step parsing approach, core source code details, practical Insert/Join examples, and extensions for lookup joins and UDTFs.
Bilibili’s FlinkSQL team has enhanced the Flink engine—now based on 1.11 with back‑ported 1.15 features—by adding Delay‑Join, table‑valued functions, projection‑push‑down, UDF and object reuse, automatic mini‑batch/two‑phase aggregation, key‑group skew fixes, connector slot‑groups, real‑time projection with Hudi, and RocksDB state‑performance tweaks, while planning remote state backends and deeper stream‑batch integration.
This article provides a detailed introduction to FlinkSQL, covering its background, the Table API, required dependencies, differences between old and Blink planners, various API usage patterns, connector configurations for CSV, Kafka, Elasticsearch, MySQL, and how to convert between DataStream and Table in Flink's unified batch‑stream processing model.
ULTRON is a one‑stop real‑time data‑warehouse development platform built on FlinkSQL that unifies data integration, asset management, cluster deployment, modeling, ETL, OLAP analysis and governance, addressing the limitations of traditional batch‑oriented warehouses and simplifying streaming data workflows for developers.