I would like to know more about how to correctly set VMX up so we can have it work within a fixed budget for overflow rendering. Occasionally our clients will pre-pay a fixed cost for additional rendering so they can receive their animation as quick as possible. In the past I have manually spooled up 250+ computers at $3/hr and watched the clock to guess where my cost is at. I know from watching Thinkbox’s youtube video (http://www.youtube.com/watch?v=pmM_8hsJt0E) that there are some budgeting controls that seem to control your burn rate ( $ per hour ) but does deadline have a way of not exceeding a fixed budget for example $20,000?
Also, where may I find tutorials and basic setup on VMX?
Thanks,
Matt Maitland
Our setup:
Deadline 7.0.0.39
Amazon EC2 c3.8xlarge Windows
V-Ray 2.40.04 & 3.0
3DS Max Design 2014 & 2015
Thank you for this and the reference image. I should clarify, I’ve used deadline monitor to spool up the 250 slaves so my tie-in to EC2 is working. I am able to get the balancer working now. I had to run the balancer on my local workstation but I suspect it is because of a technical issue with our network’s setup. (The internet is disabled on the cloud VM’s).
Thanks for asking around to see if it is something that could be done. For my studio, if we had that feature it could make things a lot more worry free and make it so we could actually sleep while things are rendering. I know Deadline doesn’t have a direct visibility on the actual cost of data, and other usage costs but I would be happy with inputting my best guess of $/hr and then have deadline track the amount of hours the instances have been running. From there it just needs to shut down the instances if the cloud’s running time reaches the specified total cost limit for the project.
A fixed-budget approach, as opposed to the existing burn-rate approach, has been frequently requested. The first release of VMX will only have the burn rate control, but we have a priority wish list item for fixed-budget control, so watch for it in a future version.
For a meantime workaround, since the Balancer algorithm is pluggable, a TD or developer could modify it to track slave statistics and cut off VM allocations at some pre-determined point. It may also be possible, depending on the provider, to force the issue by limiting re-billing through your cloud account settings.