Going 100% virtual

loocas | opinions,technical | Friday, April 22nd, 2016

AWS banner

Since february this year my studio pipeline went 100% virtual. Due to several reasons I had to get rid of my physical, dedicated render farm, disk station, rack, pretty much the entire server room, and moved to cloud computing.

It goes without saying that it wasn’t as simple and straight forward as I had hoped. Setting up everything took several days (mostly trial/error), but in the end, I moved my entire license server, render farm, Deadline etc. to Amazon’s cloud computing platform.

I picked Amazon for no particular reason. I just had the most experience with it and didn’t really look elsewhere. The setup is so complicated that I didn’t want to invest any more time and money into setting up Azure or Google cloud. The only reason would, potentially, be the cost. Because, let me tell you, virtual computing isn’t exactly cheap.

I’ll discuss my setup, my experience and my reasoning for going to the cloud in my later posts.

EC2 banner

5 Comments »

  1. Hi mate,

    I wanted to cross that bridge few years ago and after setting all up I found out that the costs would be too high once you need to render not only high ress, but previews as well.
    I’m eager to know more about all you findings.
    Please keep us posted

    Cheers

    Stan

    Comment by Stan Brusse — April 22, 2016 @ 09:08

  2. Hey, Stan,

    yes, the costs are rather substantial. While your own HW deprecates over years, virtual machines keep up with the price and therefore in a long run virtual machines are more expensive. The pros are flexibility and hassle free management. I couldn’t keep up my physical render room due to many reasons I’ll discuss in my later posts.

    Comment by loocas — April 22, 2016 @ 09:22

  3. It’s interesting how this will turn out for big company, as for me as freelancer this turned out really well :) it is really good alternative and i think best alternative to renderfarms. Used Amazon also, totally satisfied. But yeah, that involves much advanced knowledge of technical stuff than just setting up distributed rendering :) I suppose you are using deadline cloud?

    Comment by Oleg — April 22, 2016 @ 10:46

  4. I’m using “regular” Deadline + VMX, yes. But the network setup was much more complicated than on a regular physical farm.

    Comment by loocas — April 22, 2016 @ 10:50

  5. Hey man,

    This seems like a big move. Lots of tinkering and technical knowledge is rather a must to get connected to AWS cloud or any other cloud whatsoever. Connecting Deadline, network setup, integration with your pipeline…

    Plus, even though it gives you flexibility from what I’ve seen the cost is pretty steep, isn’t it?

    Comment by Lucas — April 22, 2016 @ 21:32

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress | Theme by Roy Tanck