RenderWheels is RenderStreet’s proprietary render queue management system. It was created from the start
to work in distributed, cloud-based environments, and it is constantly fine-tuned to optimize
delivery of projects of any size, as fast as possible.
FASTER DELIVERY OF ANIMATIONS
Let's look at what RenderWheels does and how it is able to achieve great scalability and
allocate render nodes in such a way so every render job is finished as fast as possible, regardless of its size. We'll
talk about animations first, as these are more resource intensive.
When an animation job is launched, RenderWheels starts rendering the first frame on one of
our servers. Once the first frame is completed, it analyzes the job parameters, the frame
completion time, and the number of frames in the job.
From there, using a complex algorithm, it dynamically allocates more nodes, to ensure a
speedy completion of the renderting. As more frames get rendered, it uses the algorithm to
refine the estimates and figure out what's needed to complete the job in a timely manner. If
needed, more servers are added or removed from the pool of servers allocated to the job.
FASTER RENDERING FOR STILL IMAGES
When talking about still images, most render farms fall into one of the following behaviors:
They either render the entire job on one server, with the drawback of taking a very long time,
or split the still image into several parts and render them on different servers. RenderWheels
does it even better.
It splits the image into parts, but instead of rendering the parts on a fixed number of servers,
it determines dynamically how many servers are needed to render the job in time. Then it
allocates the resources needed in a continuos way, so it can ensure the still images are
completed on time as well.
RenderWheels makes sure your jobs don’t wait in a rendering queue. The only delay in starting
a job is spinning up a server, if none are available when you launch it. Your jobs are always
first in line.
Using RenderWheels, jobs are completed faster and the computer power is delivered exactly when needed,
on an ongoing basis. The graphic below shows a comparison between a normal queue management
software and RenderWheels. Note how the job is finished faster as resource allocation decisions
are made on the fly.