We have just completed a transition from our old “snowflake server” installation (i.e. managed entirely manually and not well backed-up) to a “DevOps” managed site, using Ansible. What this means is that we are now able to completely rebuild this site by running a few scripts (and in fact, you are reading this on the newly-rebuilt site).
It’s not perfect: I did make a few mistakes in writing the scripts that I didn’t catch in testing on our local “staging” virtual machine. And there is a pretty serious problem with how long it takes to upload data from the management computer I’m using, because we have an “ADSL” connection with a very slow upload speed.
I have not yet made any significant changes to the site design, and have not yet attempted to roll out the TACTIC installation. In fact, I’ve run into a serious roadblock in that our present VPS service will not allow us to use the FUSE-based S3-storage solution I was going to implement to accommodate the large repository we’re going to need.
Because of this, we’re going to have to relocate our site before we can set up a lot of the expanded services I have planned. I’ve researched some different options, and it looks like the best candidate is to just switch to a “droplet” VPS from Digital Ocean. This is a more automated and less personal service than what we’ve been getting with Rose Hosting, but it is a great deal cheaper and it will allow us to set up the connected S3 storage that I had planned (we’re currently paying $40/month for our VPS, about $20 of which is paying for more disk space to accommodated the Subversion repo, whereas we can probably use the smallest “droplet” server rate of $5/month, with an additional $5-$10 going to pay for the S3 storage we’ll need).
I’m aware of several different “Infrastructure as a Service” (IaaS) providers. Digital Ocean is mainly interesting in that they have fixed VPS profiles and pricing, similar to what we’ve been doing, so it’s less of a change and the cost is predictable.
I also looked into moving to a dedicated, co-located server, which has advantages, including being able to transfer large files by simply going to the datacenter and swapping a hard drive. But with prices starting around $80/month, it’s still impractically expensive, so I don’t think we’ll go that route (yet).
Given the complexity of this transition, I’ve decided to delay migrating to a new asset management system until after we complete Episode 1. So we’ll be doing all that with the present Subversion-based workflow.
I will be applying my development effort next to getting our rendering workflow and hardware working.
