What is ORC

Definition

ORC (Optimized Row Columnar) is an open-source columnar file format designed for high performance in the Hadoop and Hive ecosystem, storing data in stripes with built-in indexes and lightweight statistics that let query engines skip data and read only what they need.

« Back to Glossary Index

Provides strong compression and efficient columnar reads for big-data analytics
Embeds lightweight indexes and statistics that enable aggressive data skipping
Integrates deeply with Hive and the broader Hadoop ecosystem
Supports complex types and ACID operations within Hive tables

Glossary: Parquet

Real World Example

A telecom running on a Hive data warehouse stores its call-detail records as ORC, using the format's built-in min/max statistics to skip entire stripes when filtering by date, dramatically reducing scan time.

FAQs

What is ORC optimised for?

High-performance columnar reads, compression, and data skipping, particularly within Hive and Hadoop-based analytics.

How does ORC differ from Parquet?

Both are columnar, but ORC originated in the Hive ecosystem and has deep Hive integration, while Parquet is more broadly adopted across engines.

What are ORC stripes?

Stripes are the large row groups ORC files are divided into, each carrying indexes and statistics that enable efficient skipping.