Apache HBase
Apache HBase is an open-source, distributed, column-oriented NoSQL database modeled after Google's Bigtable, built to run on top of the Hadoop Distributed File System and provide random, real-time read/write access to very large datasets.
Definition
Apache HBase is an open-source, distributed, column-oriented NoSQL database modeled after Google's Bigtable, built to run on top of the Hadoop Distributed File System and provide random, real-time read/write access to very large datasets.
Overview
HBase was created to address a gap in the Hadoop ecosystem: Hadoop and its distributed filesystem excel at large batch processing jobs, but they were not designed for fast, random access to individual records. HBase fills that gap by layering a sparse, distributed, sorted map on top of Hadoop's storage layer, letting applications perform low-latency lookups, inserts, and updates over datasets of billions of rows and millions of columns. Data in HBase is organized into tables made up of rows and column families, and it is stored in a sorted order by row key, which enables efficient range scans. Tables are automatically split into regions as they grow, and these regions are distributed across RegionServers, allowing HBase to scale horizontally simply by adding more nodes. Historically, HBase relied on Apache ZooKeeper for coordinating RegionServer assignment and detecting node failures. HBase is frequently used alongside other components of the Hadoop and big data ecosystem, such as Apache Spark for analytics and Apache Kafka for streaming ingestion, and it remains a common choice when an organization needs a proven, strongly consistent, wide-column store for very large, sparse datasets that must support both random access and large scans.
Key Features
- Column-family-oriented storage modeled on Google Bigtable
- Runs on top of the Hadoop Distributed File System (HDFS)
- Automatic horizontal scaling through region splitting
- Strong row-level consistency for reads and writes
- Efficient range scans due to row-key sort ordering
- Integration with the broader Hadoop and Spark ecosystem
- Support for billions of rows and millions of columns per table