Clustrix

Clustrix Inc
	File:Clustrix logo blue.png
Company type	Private
Industry	Computer database
Founded	December 15, 2006 in San Francisco, California, U.S.
Founder	Paul Mikesell, Sergei Tsarev, Eric Hoffman
Headquarters	San Francisco, CA, United States
Products	Clustrix Database Server
Number of employees	40–50
Parent	MariaDB Corporation AB
Website	clustrix.com

Clustrix, Inc. is a privately held enterprise based in San Francisco, established in 2006, specializing in the development of a database management system categorized under the NewSQL paradigm.

History

The company was founded in November 2006 with backing from Y Combinator, initiated by Paul Mikesell (previously with EMC Isilon) alongside Sergei Tsarev. Customer trials of its technology commenced in 2008.

Originally referred to as Sierra during its development phase, the platform was officially introduced to the market in 2010 under the name Clustered Database System (CDS). In December 2010, Clustrix secured $10 million in funding from investors including Sequoia Capital, U.S. Venture Partners (USVP), and ATA Ventures. Robin Purohit, who had experience at Hewlett-Packard, assumed the CEO role in October 2011. Additional financing rounds followed: $6.75 million raised in July 2012, $16.5 million in May 2013 from existing stakeholders, and a $10 million injection led by HighBAR Ventures in August 2013.

Mike Azevedo took over as CEO in 2014, succeeding Purohit. By February 2016, the company had accumulated over $23 million through debt funding. Clustrix’s innovative database technology and strategic direction culminated in its acquisition by MariaDB Corporation, an announcement made on September 20, 2018.

Technology

Clustrix functions as a scalable solution for transactional processing and real-time analytics that can be implemented as a seamless substitute for MySQL, overcoming MySQL’s limitations regarding scalability with minimal operational impact. The system incorporates inherent fault tolerance to ensure high availability and supports parallel backup and replication mechanisms for robust disaster recovery.

As a horizontally scalable SQL database, Clustrix aligns with the NewSQL initiative by modernizing relational database management systems through distributed system architecture. Initially distributed as a hardware "appliance" leveraging InfiniBand connectivity until approximately 2014, the product transitioned to downloadable software and became accessible via the Amazon Web Services Marketplace by 2013.

Clustrix distinguishes itself from major competitors such as Microsoft SQL Server and MySQL by delivering a distributed, ACID-compliant relational database capable of scaling transactional workloads while also facilitating real-time analytics. Unlike column-oriented distributed databases like EMC Greenplum, HP Vertica, and Amazon Redshift, which prioritize offline analytical processing, Clustrix is optimized for transactional use cases. In-memory databases such as VoltDB and MemSQL provide excellent performance for low-latency transactions but do not support real-time analytics functionality. Conversely, NoSQL databases like MongoDB excel at managing unstructured data and read-heavy workloads but lack support for transactional write-intensive operations. Consequently, NewSQL and NoSQL solutions serve complementary roles within the database ecosystem.

Query evaluation

The Clustrix system operates on a distributed cluster composed of shared-nothing nodes, utilizing a query-to-data methodology. Each node manages a specific portion of the dataset, and SQL queries are decomposed into segments that are dispatched to the appropriate nodes. This design facilitates horizontal scalability by simply adding more nodes.

Data distribution

Data within Clustrix is automatically partitioned and evenly spread across all nodes, ensuring balanced distribution even as nodes join or leave the cluster. This approach removes the necessity for manual sharding and guarantees strong availability in the event of node failures.

Performance

According to a 2011 Percona benchmark, a three-node Clustrix cluster delivered approximately 73% better performance under heavy workloads compared to a single MySQL server. Performance improvements scaled nearly linearly with the addition of more nodes, highlighting the effectiveness of its scalable infrastructure.

Project cancellation

MariaDB declared in October 2023 that it would discontinue Xpand (previously known as Clustrix) following a corporate restructuring initiative.

External links