spark-plan-viz 1.1.0

A tool to visualize Apache Spark execution plans.

Spark Plan Viz

A lightweight, interactive tool to visualize and analyze PySpark execution plans using D3.js. It helps developers and data engineers debug complex queries when the textual result of df.explain() is not enough.

Documentation | PyPI | Optimization Reference

Features

• Interactive Visualization: Zoom, pan, and click nodes to explore execution details
• 14-Rule Optimization Engine: Detects cross joins, missing broadcasts, full table scans, Python UDFs, and more
• Performance Insights: Instantly identify shuffles, broadcast joins, and pushed filters
• Jupyter Integration: Renders directly inside notebooks without external files
• Standalone HTML: Export visualizations to share with your team
• AQE Support: Full support for Adaptive Query Execution details

Installation

uv add spark-plan-viz

Or with pip:

pip install spark-plan-viz

Quick Start

Visualize

from spark_plan_viz import visualize_plan

# Renders inline in Jupyter
visualize_plan(df, notebook=True)

# Or export to HTML file
visualize_plan(df, notebook=False, output_file="my_plan.html")

Analyze

from spark_plan_viz import analyze_plan

suggestions = analyze_plan(df)
for s in suggestions:
    print(f"[{s.severity.value}] {s.title}: {s.message}")

Example

from pyspark.sql import SparkSession
from pyspark.sql.functions import broadcast
from spark_plan_viz import visualize_plan

spark = SparkSession.builder.appName("Example").getOrCreate()

orders = spark.createDataFrame([
    (1, "Alice", 100),
    (2, "Bob", 200),
], ["id", "customer", "amount"])

customers = spark.createDataFrame([
    ("Alice", "NY"),
    ("Bob", "CA"),
], ["name", "state"])

result = orders.filter(orders.amount > 50) \
    .join(broadcast(customers), orders.customer == customers.name) \
    .groupBy("state") \
    .agg({"amount": "sum"})

visualize_plan(result, notebook=True)

What the Optimizer Detects

The built-in analyzer checks for 14 common performance issues:

Severity	Rule	What it catches
ERROR	Cross Join	df.crossJoin(other) — Cartesian product
ERROR	Nested Loop Join	Non-equality join condition → O(n*m)
WARNING	Full Table Scan	Scan with no pushed filters
WARNING	Expensive Collect	collect_list / collect_set in aggregates
WARNING	Window Without PARTITION BY	Global window → single partition
WARNING	Python UDF	Row-level JVM↔Python serialization
WARNING	Redundant Shuffle	Back-to-back repartition() calls
WARNING	Partition Count	< 2 or > 10 000 partitions
WARNING	Sort Before Shuffle	Sort immediately destroyed by shuffle
INFO	Missing Broadcast Hint	Shuffle join where broadcast may help
INFO	Non-Columnar Format	CSV/JSON scan instead of Parquet/ORC
INFO	Coalesce via Round-Robin	repartition(n) when coalesce(n) suffices
INFO	Skew Hint	SortMergeJoin without skew optimization
INFO	Unnecessary Sort	Sort not consumed by ordering-dependent op

# Quick example: detect a cross join
result = employees.crossJoin(departments)
suggestions = analyze_plan(result)
# → [ERROR] Cross Join Detected: Cross joins produce the Cartesian product ...

# Fix it:
result = employees.join(departments, employees.department == departments.dept_name)
suggestions = analyze_plan(result)
# → no cross_join finding

See the Gallery for runnable examples of every rule with bad patterns and fixes.

Requirements

• Python 3.11+
• PySpark 3.x+
• For notebook mode: IPython/Jupyter

Limitations

spark_plan_viz is not available on Databricks Serverless Compute, as it's not possible to access the _jdf object of a Spark DataFrame.

License

MIT License

Contributing

Contributions are welcome! See CONTRIBUTING.md for guidelines.

Development Setup

git clone https://github.com/montanarograziano/spark_plan_viz.git
cd spark_plan_viz
just install
just pre-commit
just check
just test

Acknowledgments

Built with D3.js for interactive visualizations.