-
After Effects Workgroup Hardware
Hey All…
We’ve been a small working group, primarily in Motion Graphics apps like AE.
Right now I have a library of assets on a server and artists working on projects primarily on local storage.
I want to move actual projects to the server side so it’s easier for any artist to pickup any projects without having to collect assets to move it….essentially, I want an edit in place workflow.
I’m not unfamiliar with this as I have use solutions from Terrablock in the past at more “editorial” oriented facilities.
The other thing I want is scalability of storage capacity. I’ve been looking at virtualization options like Clustore or Poolit. This will allow consistent mount points even when expanding capacity, and not tie to a particular architectures.
Outside of that….What a hardware are ppl using? As we are comping and not editing and, what’s the measure of performance? Editorial tends to base things on stream count. But comps are layers that need to pull lots of media….so what’s a good metric? We are equipped with local SSD’s for fast caching while working.
What raid configurations? Small Tree and Tiger Technologies come to mind.
We are inside of a larger organization, should we consider secondary HBA’s or NICs and put ourselves on an isolated groups for media traffic?
Just trying to get a discussion started….or recommendations…thanks!
Alex Udell
Editing, Motion Graphics, and Visual FX -
1 Reply
-
Hi Alex,
You probably know more about this than I do (but hide it better :)) – I just want to follow this discussion and see what comes out of it, and perhaps ask a few questions…
[Alex Udell] “As we are comping and not editing and, what’s the measure of performance?”
The only fool-proof measure I know of, is matching current and anticipated demand – so somehow, however difficult it might be, gotta measure that demand. Other than that, it’s throughput, I/Os and latencies (the latter two being related) – can’t think of anything else.
Have you tried figuring out the demand for I/Os and throughput – per client and aggregated?
[Alex Udell] “What raid configurations?”
Why, RAID10 of course… It’s a cure for all ills… Kidding! 🙂 Depends on quite a few factors, not the least – desired performance. Most popular options in single tier storage are 6, 10, and single or dual parity variants of RAIDZ. With a high demand for I/Os and low latencies, RAID10 comes to mind – but it’s relatively inefficient; if 6 works for you, I’d choose 6.
[Alex Udell] “We are inside of a larger organization, should we consider secondary HBA’s or NICs and put ourselves on an isolated groups for media traffic?”
Probably the best option; subnets on VLANs and/or dedicated hardware would also work, depending on the configuration. The idea is to ensure that those local packets get routed locally rather than through the main rack. If there is an option to talk to your local IT people and ask what they would recommend to ensure local routing for your network segment, I’d do that first.
Alex Gerulaitis
Systems Engineer
DV411 – Los Angeles, CA
Reply to this Discussion! Login or Sign Up