WebPagetest Forums

Full Version: Question about EC2 Server AMI - AWS Cost per Hour
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Hello,

I am planning a migration of our current private instance to AWS using the pre-built AMIs and am having a problem finding out exactly what cost per hour the ami-fcfd6194 server would be. Is this image based on the cheaper Amazon Linux (Ubuntu version) that is currently shown to cost $.108 per hour for an m4.large for on-demand pricing? I've reached out to an Amazon Solution Architect, and he was not even sure of the specifics of that particular AMI. Would anyone happen to know exactly which costs structure would be used for the following?

ami-fcfd6194
m4.large
On-Demand Pricing for now.

Thanks,

Pete
I don't see that AMI in the list anywhere: https://sites.google.com/a/webpagetest.o...-Virginia-

Is that a copy you made of one of the public images?
Sorry, just realized that is the server image and not the agents.
The server image is Ubuntu-based but from 2014. AFAIK there are only 2 tiers for pricing - Windows or Linux and this would definitely be the cheaper Linux pricing but if you have different tiers beyond that I can't really help.

In case it wasn't obvious, the server doesn't actually run any tests and it will still need to start/stop agents and the agents will get billed at the windows rate (the server doesn't know anything about the new Linux agents yet).
Just checked the pricing sheets and yes, it is the $0.108/hr Linux pricing for the server.
Awesome! You just made my day!

Regarding the other comments, yep we will have a bunch of the Windows agents driven via API (probably have to have some dummy tests to keep them up all the time) but we actually will be able to expand by almost double and keep the costs lower than our Azure setup (and have things more consistent as our machines were all over the place)

Thanks for the quick replies, and for the info on the Server AMI/New Agent incompatibility. The concept behind those is to spin up one cheap Linux machine and run docker containers within that, effectively boosting the throughput of pages per hour per Linux VM? If so, sounds like a great stride forward. A little hesitant to entrust things to Docker right now (mainly my lack of Docker prowess), but yeah maybe next year Smile

Thanks again,

Pete
could you not run this on something smaller? m4.large is pretty meaty. I am intrigued at your end solution, as we got a bit of a shock when the Antipodean node disappeared for a few days.

( on another note, would it be possible to donate regularly to support this public node?? )
Reference URL's