SOURCE: Citus Data

Citus Data

March 26, 2015 09:18 ET

Citus Data Releases New Version of CitusDB 4.0 for Hybrid Transaction/Analytical Processing on PostgreSQL 9.4

pg_shard 1.1 and cstore_fdw 1.2 Open Source Extensions Also Released

SAN FRANCISCO, CA--(Marketwired - Mar 26, 2015) - Citus Data, creators of solutions to scale out and extend the analytics capabilities of PostgreSQL, today announced the general availability of CitusDB 4.0, the latest release of the company's hybrid transaction/analytical processing platform for massively scalable, high availability PostgreSQL. CitusDB 4.0, based on the latest PostgreSQL 9.4 release, enables companies to massively scale out PostgreSQL across commodity servers and use parallel processing to distribute queries across the cluster for real-time analytics queries on big data. The CitusDB 4.0 release improves performance and usability while migrating to the PostgreSQL 9.4 platform, which provides greatly enhanced support for unstructured data types. In conjunction with the release of CitusDB 4.0, Citus Data released new versions of the company's open source projects, pg_shard, and cstore_fdw.

CitusDB 4.0
The advanced CitusDB platform powers hybrid transaction/analytical processing across hundreds of billions of events while cutting query times by up to 100x. By moving to PostgreSQL 9.4, the platform now provides enhanced support for both structured and unstructured data through JSONB. CitusDB 4.0 brings exciting new functionality and performance improvements, including:

  • Support for real-time workloads for hybrid transaction/analytical processing - CitusDB Enterprise Edition now integrates with pg_shard to support real-time workloads. This integration combines scalable analytics and low-latency writes in a single system.
  • Integration with PostgreSQL 9.4 - New features include JSONB support, a faster, more efficient data type for storing JSON data, faster and smaller GIN indexes, and more PostgreSQL 9.4 features.
  • Ability to re-balance the cluster for incremental scalability and fault tolerance - Users can incrementally add nodes and uniformly distribute data, and thus traffic, to those nodes. Users can also re-replicate data from failed nodes evenly across all the remaining nodes.
  • Faster query performance - A new task-assignment policy provides better in-memory workload performance, while binary serialization for data copied between nodes provides faster performance on queries fetching a lot of data. Batching task-assignment calls improves performance with re-partition joins.
  • Improved usability - Modified \STAGE allows loading data from any node in the cluster, making data-loads much easier and allowing for a more uniform data placement. Query throttling in the real-time executor prevents resource exhaustion when queries touch thousands of shards.

pg_shard 1.1
pg_shard is an open source extension for transparently scaling out PostgreSQL horizontally across commodity servers. pg_shard automatically shards Postgres tables and replicates the shards on multiple nodes in the cluster, ensuring high availability for PostgreSQL databases. Updates in the current release include:

  • Improved performance - The INSERT commands run up to four times faster.
  • Shard repair - Inactive placements can be easily brought back up to speed.
  • Copy script - Data from .csv and other files can be quickly imported from the command line.
  • CitusDB integration - pg_shard's metadata can now be exposed for use by CitusDB.
  • Resource improvements - Larger queries can be executed.

pg_shard 1.1 can be downloaded from GitHub at https://github.com/citusdata/pg_shard.

cstore_fdw 1.2
cstore_fdw (PostgreSQL Columnar Store Extension) is an open source extension for creating columnar PostgreSQL tables, which can reduce query times as well as storage footprint. Updates in the current release include:

  • Support for "INSERT INTO ... SELECT ..." - Data can now be loaded directly into cstore from regular PostgreSQL tables.
  • Support for "COPY TO" - The COPY command can now be used to copy the contents of a cstore table to a file.
  • Improved Memory Usage - As much as 90 percent less memory is used in some scenarios.

cstore_fdw 1.2 can be downloaded from GitHub at https://github.com/citusdata/cstore_fdw.

Quotes

Umur Cubukcu, CEO and co-founder of Citus Data
"The Citus Data product family of CitusDB, pg_shard, and cstore_fdw allow companies to use PostgreSQL to tackle large datasets and big data analytics challenges. Whether companies are facing scaling issues, need to reduce query times and storage footprint, or want to simplify their transaction and analytical database infrastructure, Citus Data products provide powerful solutions. We are excited about the release of CitusDB 4.0 which now seamlessly integrates with pg_shard on PostgreSQL 9.4 to deliver a very powerful solution for companies that want to slash the cost and complexity of their database infrastructure by implementing a single database for both their transaction and analytical processing."

About Citus Data

Citus Data offers solutions which power massively scalable PostgreSQL for real-time applications. The CitusDB hybrid transaction/analytical processing platform powers real-time workloads across hundreds of billions of events while cutting query times up to 100x and supporting both structured and JSON data. Citus Data also offers pg_shard, an open source extension for scaling out PostgreSQL across commodity servers, and cstore_fdw, an open source extension for creating columnar PostgreSQL tables. Citus Data is a Y Combinator alumnus, and is backed by investors that include Data Collective, Bullpen Capital and SV Angel. All Citus Data products are available for download at www.citusdata.com.

Contact Information

  • Press Contacts
    Terry Erisman
    Citus Data
    (415) 688-4093
    Email Contact