Just want to speak about a research project I’m working with.
It’s a deployment framework to launch blender clusters in Public/Private
clouds (Currently AWS is the only supported provider).
We have worked with it off-line with a few clients and we just launched
recently a public version.
It’s called RendIt (www.rendit.org), it allows to orchestrate
the deployment of on-demand render clouds, it’s an automation
framework to deliver Blender render farms in only 3 steps
using your AWS credentials and almost without any effort.
It’s especially useful when using with AWS free-tier as we will
have deployed freely the infrastructure and the orchestration services,
a win-win relationship.
We support render multiple frames or single frame
scenes in multiple nodes.
It’s for you the thanks, not us! Great news, Friend! I know very well AWS and your job is very interesting for everyone who use Blender!
At soon…
Spirou4D
Thanks a lot for the feedback, much apreciated. I got some questions related to the platform like why is not OpenSource or how can we check locally what’s going on in the AWS nodes.
I’m preparing some answers to post it here, but, just as an intro even if the web app is not OpenSource (yet) what is deployed at the end is pure Python using
bpy. We don’t persist any data (We don’t have a DB), so, we don’t store the private keys to access users servers (This is because the instances should live only when doing the render and after this, users should remove the job removing the cluster), in the case a user needs to have access to the data inside the nodes you can mount the volume in another instance and check it without issues.
We don’t persist any data (We don’t have a DB), so, we don’t store the private keys to access users servers (This is because the instances should live only when doing the render and after this, users should remove the job removing the cluster),
yes it’s secure so!
And if it’s a lot of time’s use to re-write our private keys, the web navigator can store our private keys in memory -> not a problem!
Thks a lot and good luck!
What you said it’s a fact, AWS doesn’t have any Free Tier for GPU capable instances.
We launched a few benchmarks and depending on the instance that you are using the time difference between a GPU instance (p2.xlarge) and the Free Tier instance (t2.micro) it’s about 15-20 times faster, but the cost of a p2.xlarge is 75 times more expensive.
Let me speak about the benefits of using the free tier instances for all users and developers, It allows us to have a testing infrastructure at 0 cost. And allow users to instead of using 1 GPU capable instance, deploy 20 smaller instances at cost 0.
This platform is used partially for academic purposes, we have some schools playing with blender, but their machines are kind of blocked when doing the render, so they launch their projects using RendIt and AWS without any issue (cost).
And you always have the ability to create more powerful instances on demand for specific tasks (And after using them you just remove the job deleting all the AWS resources created).
We are running some experimental jobs for using luxray, maxwell and vray and they will be released in the platform this quarter. There are a few licensing issues we are facing but hopefully, we will be able to integrate commercial/licensed render engines.