| RenderFarmer Menu: |
RenderFarmer Studio
for Render Farm Creation
Our software makes render farm creation easy for anyone who can install Linux. The solution stack includes Centos Linux 5.x, the DrQueue open source Distributed Render Queue, and our very own ClusterMaker Linux cluster creation software.
All you need is a Linux system running Centos 5. Then you can register for a free trial download of RenderFarmer and you'll get access to our download area. There you'll find links to our latest development releases for RenderFarmer and complete technical support.
How does it work?
RenderFarmer is a suite of software that can be up and running about 40 minutes after beginning our automated installation routine. The suite will set up a Master Server to control your new cluster and will provide a set of tools for you to add and remove render farm nodes and modify the configuration. The intuitive web interface make is easy to add nodes and monitor performance with point and click ease, but every graphical tool has a command line equivalent and the cluster subsystems can be operated via ssh.
You will have to install your render engine (maya, blender, etc.) on the Master Server and add some render nodes, which is as simple as setting a desktop computer or server for PXE boot in the BIOS, connecting it to the network, and then turning it on. After it attempts to boot, just log in to the Master Server's web interface and choose Node Management, Add Node. The computer you just powered on will appear in the menu of available nodes and you will be prompted to give it a name and an IP address, and that's it! Click Finish. What other vendor can talk you through scaling up your render farm from bare metal to online in three sentences?
Ease of use
That kind of ease of use and deployment speed is unheard of in the render farm management space. After a reboot, your new render farm node will be waiting for render jobs. Up to ten nodes can be added at a time. For a more detailed overview, read our article discussing how to build a render farm with our software.
It may be possible scale quickly if you have a custom provisioning system already in place, but even then you are deploying images to a disk where they will later need to be maintained. Also, hard drives are the most common failure in typpical nodes which further adds to your maintenance burden. But in our render farm management software, everything runs 100% in memory on our PXE booting compute nodes. Zero maintenance, diskless compute nodes. They can be added, removed, updated, and swapped out in seconds. Big job coming up? You can add 20 nodes in a few minutes, and then remove them just as easily when you're done without having touched their hard drives. This means they go right back to their "prior life" without any reprovisioning.
Extreme Scalability
Among other things, we have built in storage scalability. Our cluster storage is based on GlusterFS, an extremely flexible cluster filesystem that can use ethernet or Infiniband as an interconnect. While the default configuration uses one server and many clients over standard ethernet, this can easily be adapted for your special needs. More servers can be added that will increase aggregate throughput, provide filesystem redundancy, increase the total available storage, or all three. For large clusters or huge render jobs, this will provide the i/o scalability you need to get the job done. Over 20 gigabytes per second of aggregate throughput have been demonstrated.
Among other things, we have built in storage scalability. Our cluster storage is based on GlusterFS, an extremely flexible cluster filesystem that can use ethernet or Infiniband as an interconnect. While the default configuration uses one server and many clients over standard ethernet, this can easily be adapted for your special needs. More servers can be added that will increase aggregate throughput, provide filesystem redundancy, increase the total available storage, or all three. For large clusters or huge render jobs, this will provide the i/o scalability you need to get the job done. Over 20 gigabytes per second of aggregate throughput have been demonstrated.