I’ve found that sometimes in the VFX industry, technology is marketed to sound great in theory but ends up failing in practice. There are times where I hear people casually mention how simple things would be if they used their idea for how technology should be organized at a facility. Now I usually tend to stay out of those conversations because they end up feeling like a pissing contest but allow me to give some examples.
Usually during the craziness of a show diskspace will start filling up and artists are asked to submit files for backup and removal. I always find myself sitting next to an artist who gripes:
Oh my God. What the hell is wrong with this place. Diskspace is so cheap. Why can’t they just buy more. Hell I should bring my 1 TB hard drive from home.
Again it’s an absurd thing to say but you always run into these casual technological opinions to major problems in the VFX industry. As one commenter said:
Thanks to silicon valley’s massive push for investment in world wide internet bandwidth during the dot com boom in the late 90s, outsourcing knowledge jobs to local people doesn’t make business sense. (Share holders want growth) Knowledge jobs are moving to the cloud. The only jobs that can’t be outsourced are the ones that can’t be done online. This results in a massive re-balancing of world wide knowledge labour markets.
Ah yes it’s so simple with cloud computing! I’ve had a few encounters with cloud computing. A company I worked for announced they had these new render procs in some other part of the country. While these procs were very fast, they had to have my file and all the geometry and textures associated with it transfered to the location of where the procs exist. Say hello to data latency: The amount of time delay for the file to make it to the location of the proc. I’ve found that this sometimes took longer than the render! It’s like having a powerful gun with a complicated reloading system.
You’d be surprised to know that many companies in the VFX industry are divided over cloud computing:
It seems like a no brainer — but not everyone is biting.
Cloud computing critics argue that many of the security systems aren’t enough to keep a smart, motivated data thief out of a client’s files. They also believe data latency — the time it takes to access off-site files, especially when those files are large — slows down the work flow too much for off-site cloud computing to be practical.
I think it’s great to see networks getting faster but what we miss is that the file sizes are getting larger also.
Fallacies of Distributed Computing
I came across this paper about many of the fallacies involved on distributed computing that should be read by every person in the VFX industry. Here are the 8 common fallacies of distributed computing:
1. The network is reliable.
2. Latency is zero.
3. Bandwidth is infinite.
4. The network is secure.
5. Topology doesn’t change.
6. There is one administrator.
7. Transport cost is zero.
8. The network is homogeneous.
I think some of us see something work in the simple test case and then just assume it will work for a vfx production. Well if you were able to do it for 20 shots why can’t you do it for 1500 right? If you think that’s possible your money will probably end up in the same place as your data: In the clouds.