What does this tool do?

The Flink SQL Playground parses your query client-side and extracts its structure: source topics, transformations, window definitions, join types, aggregations, and output. No Flink cluster required — everything runs in your browser.

Topic flow diagram

See how data flows through your pipeline — from Kafka source topics through filters, joins, windows, and aggregations to the output stream or table. Each operation is color-coded by type.

Supported Flink SQL features

Window functions

TUMBLE, HOP (sliding), SESSION, and CUMULATE windows — both legacy GROUP BY syntax and modern TABLE-valued function (TVF) syntax. Visual timeline shows how events are grouped.

Join types

Inner, left, right, and cross joins. Temporal joins with FOR SYSTEM_TIME AS OF for versioned table lookups. Interval joins with BETWEEN for time-bounded stream-to-stream correlation.

Complex Event Processing

MATCH_RECOGNIZE pattern detection — fraud detection, price trends, sequence matching. The tool extracts PATTERN definitions and DEFINE conditions.

State analysis

Automatic risk assessment — low (windowed, bounded), medium (temporal join, session window), or high (unbounded GROUP BY, MATCH_RECOGNIZE). Helps prevent state explosion in production.

Confluent AI/ML

ML_PREDICT, AI_COMPLETE, AI_EMBEDDING, VECTOR_SEARCH_AGG, and other Confluent Cloud for Apache Flink functions. Detects model references and function types.

Analytics & dedup

ROW_NUMBER, LAG, LEAD, FIRST_VALUE, LAST_VALUE with OVER clauses. Detects the common deduplication pattern (WHERE rn = 1).

Who is this for?

Data engineers building Flink SQL pipelines on Apache Flink or Confluent Cloud. Useful during development to understand query structure, validate window semantics, and catch unbounded state issues before deploying to production.