Our software is designed to enormously simplify the use of advanced Linux clusters, whether for render farms or other distributed processing. Since the beginning of Linux in the data center, the open source community has been producing outstanding cluster tools. However, they were each designed with different goals in mind and can be difficult to integrate and configure.

Customers asked for Linux clusters that met the same basic requirements:

• Easy to install and configure
• Rapidly scalable
• Fault tolerant
• Able to share a workload
• Energy efficient

Over the last several years, we integrated and tested many of the most popular open source cluster tools and worked on stabilizing a good configuration. Then we wrote a graphical web interface and command line tools that would make controlling this array of software much easier. Now anyone can install a render farm or load balancing Linux cluster in about 40 minutes and put their compute nodes right to work.

After ease of use, rapid scalability was our other mandatory design goal and our shared root clusters are designed to seamlessly grow and shrink according to demand. If you have a render farm and want to scale up for a big job, it's never been this easy.

Imagine no provisioning… Imagine no software installations… Just PXE boot some extra hardware, ANY hardware, and you're up and running with more render nodes!

Not only can you scale in seconds, but since you never “provisioned” the hardware, you don't have to “provision it back” to what it was doing before. The PXE boot never changed what was on disk, so a reboot will return your extra render nodes to their previous state as workstations, spare hardware, or whatever else.

In another scenario, you might have a Linux cluster providing web services. It can now grow in seconds or minutes to meet increasing demand, and it has internal IP load balancing. No need for an external load balancer costing thousands of dollars! Likewise, in times of low demand it should be able to shrink and conserve significant power and cooling costs. No longer does a cluster need to be statically configured to meet peak demand.

This alone saves enough energy to eventually pay for the software… Consider that at current electricity prices, each 500 watt server consumes about $1000 per year in electrical and cooling costs when it's on 24 hours a day. If your 5 node web application cluster only needs all of the cluster nodes during a peak period and can leave 3 of them off for the other 18 hrs per day, that's a $2250 annual savings! As data centers are pressed to reduce energy consumption and become as “green” as possible, this is a huge step in the right direction.

Note: While the software design is inherently distribution agnostic, at this time it works only on RedHat Enterprise Linux 4.x / 5.x and CENTOS 4.x / 5.x. We recommend using 5.x wherever possible as support for 4.x will eventually be discontinued. We are working on porting the software to Debian / Ubuntu and other Linux distributions according to your votes (vote in the poll on the right!).