Apache HBase
| Company type | Open-source software |
|---|---|
| Industry | Database management system |
| Founded | 2010 |
| Products | Distributed non-relational database |
| Parent | Apache Software Foundation |
| Website | hbase |
HBase is a Java-based, open-source distributed non-relational database inspired by Google's Bigtable. It is developed as part of the Apache Hadoop ecosystem under the Apache Software Foundation, functioning atop the Hadoop Distributed File System (HDFS) or Alluxio. HBase provides Bigtable-like functionality within the Hadoop framework, offering a resilient method to store vast amounts of sparse data and enabling rapid access to critical elements in extensive datasets.
Among its principal capabilities are data compression, in-memory processing, and per-column Bloom filters, reflecting the original Bigtable architecture. HBase tables work natively with Hadoop's MapReduce workflows and can be accessed through Java APIs or via REST, Avro, and Thrift interfaces. As a wide-column store tightly integrated with Hadoop and HDFS, it supports high-volume, low-latency read and write operations on large-scale data.
Though not a direct substitute for traditional SQL databases, the Apache Phoenix project introduces a SQL interface and JDBC driver to facilitate HBase’s use with analytics and BI platforms. Furthermore, Apache Trafodion provides a SQL query engine with ODBC/JDBC support and distributed ACID transactions spanning multiple statements and tables, using HBase as its storage backend.
HBase has been employed by various data-centric websites; however, Facebook transitioned its Messaging Platform from HBase to MyRocks in 2018. Unlike conventional relational databases, HBase lacks native SQL scripting—comparable tasks are implemented in Java, similar to MapReduce programming.
In line with Eric Brewer’s CAP Theorem, HBase is identified as a CP system, emphasizing Consistency and Partition tolerance.
History
The inception of Apache HBase traces back to Powerset, where it was created to manage large-scale natural language search data. It achieved top-level Apache project status in 2010.
Facebook initially adopted HBase for its messaging infrastructure in November 2010 but completely phased it out by 2018.
The current stable version belongs to the 2.4.x release series, superseding previous editions.
Enterprises that use HBase
Prominent organizations utilizing HBase include:
- 23andMe
- Adobe
- Airbnb, employing HBase within its AirStream real-time analytics platform
- Alibaba Group
- Amadeus IT Group, serving as their main long-term storage solution
- Bloomberg, for managing time series datasets
- Facebook (2010–2018)
- Flipkart, applied for search indexing and customer insights
- Flurry
- HubSpot
- Imgur, powering notification services
- Kakao
- Netflix
- Quicken Loans
- Rocket Fuel
- Salesforce.com
- Sears
- Sophos, supporting backend operations
- Spotify, utilized for Hadoop integration and machine learning tasks
- Tuenti, used in messaging platforms
- Xiaomi
- Yahoo!
See also
- NoSQL
- Wide column store
- Bigtable
- Apache Cassandra
- Oracle NOSQL
- Hypertable
- Apache Accumulo
- MongoDB
- Project Voldemort
- Riak
- Sqoop
- Elasticsearch
- Apache Phoenix
Bibliography