This is a business question, which I find hard to answer because I will never be as familiar with your business plan as you are.
I've posted our pricing information in the other issue, but here is the important bit:
We charge you 150$ per server we manage.
You might be able to get about 100 small sites on 1 of our servers
This is highly variable and needs stress testing
Basic calculations based on 20$/month:
If you can get only 20 sites on one server, we'll charge you about 320$ and you will be charging about 400$
50 sites: Costs you 620$, gives you 1000$
100 sites: Costs you 1150$, gives you 2000$
200 sites (optimistic): Costs you 2120$, gives you 4000$
This is not counting any additional support costs. As you can see, the profit margin more or less (logically) flatlines after 50 sites, but it's my educated guess that having more sites per server will lead to unforseen management issues.
(Anything past this seems like speculation to me, and that's just not my field :) )
There's no bandwidth limit. The storage space is 50GB. We haven't moved you yet since you're planning on moving to AWS, but normally we would transition you to the VPS cloud 2, which is the same as you have now minus 10GB of disk space: https://www.ovh.com/ca/en/vps/vps-cloud.xml
We can upgrade a server to VPS cloud 3 if needed, with a proportionnal raise in fees.
I don't know if you and Christian have had the conversation regarding rates but most of our fees are for the service so it doesn't matter much where the server is in terms of pricing.
Would that be a separate setup/instance from my AegirVPS? Just wondering how this is different if we're providing allOpenProducer hosting on AWS.
What do you mean? We would charge you 150$ per server we manage, and we only manage Aegir servers. I expect you can get about 100 small sites on a server (on our servers), though this is highly variable and there's obviously no way to know (yet) for your specific distribution. We charge 10$ per site (-3 free sites), so a full server would cost you about 1120$.
I don't know how many sites an AWS t2.micro instance can handle, presumably it's less.
Designing a different architecture would be a rather ambitious project, though for the usual price I could route them through Varnish or nginx (to have more than one server using the same *.openproducer.com wildcard).
Basically you can put the makefiles anywhere, I put them in /var/aegir/makefiles/[project-name] as a convention.
From there you should be able to use the drush makefile when creating a new platform, you don't need to create it with drush make in advance.
I hope this helps.
I added the automated OpenAtrium platforms. Keep in mind these are directly Phase2's makefiles, so we can't do much testing and are dependent on their QA. (After 45 versions though, I would hope they know what they are doing.)
So you (usually) have to clone it with Git to use it. You usually have to read the README to know which makefile is the "right" one if there is more than one. There is a sort-of-convention emerging that has people naming them build-[project].make , probably something to do with how drupal.org tests them. In this case build-redhen_raiser.make seems right.