About Us


The need for a scale-out database

After starting Isilon, Paul Mikesell knew there was a huge need for scale-out infrastructure. With his experience as a programmer, Paul knew that the next problem he wanted to solve for was a scale-out database. At this same time, Sergei Tsarev had built a time-series database for AOL and independently had the same idea to build a database that could handle transactions and also scale, and was part of YC ‘06. Sequoia was interested in funding the scale-out database idea, but told Paul and Sergei that they would only fund it once and the two met with one another. Paul and Sergei scheduled to meet over coffee and were surprised to find that they actually got along. Long story short, together Paul and Sergei started Clustrix!

From the ground up

Starting things the hard way, Paul and Sergei started building a scalable database from scratch (and not just creating a bolt-on solution to an existing database the way many others have). They decided to be MySQL compatible, but leverage distributed query compilation and execution, and created novel ways of distributing and replicating data (which are now patented).

Spinning disks

In order to build a high-performance distributed database, they also knew they needed SSDs. Back in 2006, SSDs were neither as mature as they are today, nor were they as durable as they claimed. Clustrix partnered with some prominent SSD manufacturers to profile behavior of disks in the face of power failure, then built a custom, battery-backed NVRAM card to buffer writes. To help ensure network traffic scaled as cluster sizes grew, they also built in support for Infiniband.

The appliance

The result was the Clustrix Database Appliance, which started shipping in 2008. It was deployed in dozens of production clusters all around the world and showed that not only was there a real need for a scale-out SQL database, but that ClustrixDB could handle those workloads.

The evolution of software

As Clustrix grew, SSDs were improving and data centers were moving to the cloud. It became clear that ClustrixDB should be available as a standalone software offering that could be deployed on commodity hardware. ClustrixDB 5.0 was the first software release that became available in 2014. Clustrix stopped selling the appliance shortly thereafter.

Today, ClustrixDB is deployed all over the world, in data centers on bare metal and in the cloud (including available as an AMI) and continues to help customers scale their OLTP workloads simply while providing high availability without sharding or code changes.