Am I missing something? Pricing seems to be at the machine level via core/memory/etc. But I don’t see anywhere to configure/create/spin up machines. I would have thought it would be in the fleet, but doesn’t look like anything is there for resource configuration.
I only see how I can set usage settings on a build configuration for density purposes, but not sure how I create a machine with X cores and Y memory to run those servers on.
So is this just me, and I’m dumb?
I only see two reports.
One is a cost & usage report that tells me how many core/memory hours I’ve used and the charge for that, per day. Okay, that’s good, but no idea how CPU cores are calculated in relation to my server CPU Mhz usage (only CPU data I can see/set, but they are two completely different metrics).
The other is at an individual server level which just tells me memory usage and CPU usage in Mhz (how does that translate to cost?) per 20 minute interval (assume averaged, or is it max?, the graph doesn’t say).
The second graph is basically useless from a cost perspective since it’s per server. How do I see that across a fleet at least? How does that equate to cost?
Not sure what I’m missing, but right now from a business cost perspective, this is completely unusable.
Hey whilke,
We currently utilize 2 core machines with 8GB of memory by default. Your usage settings will determine how many servers will fit on one of these machines. If you need more servers than can fit on a single machine per those usage settings another machine will be brought online in these 2 core CPU/8GB memory increments to meet your demand. The per server CPU and memory statistics will give you a baseline to configure your usage settings to in order to optimize your server per machine density which will also optimize your costs in turn.
How does the 2 cores map to server CPU MHz settings for density? I assume the process has access to both cores?
i.e. what’s the math to compute servers per machine as it relates to cores to mhz?
Each core has a base 2.8GHz by default. The server process does indeed have access to both cores for a total of 5.6GHz available compute power per machine. Your server density is determined by dividing your defined usage cap (set in the build configuration) into that total. The same occurs for your defined memory usage resulting in packing the maximum number of server instances per machine as those two usage settings allow.
Thanks. That is the info I needed. I hope you guys are updating your documentation to include this.
This information should be added to the documentation. I was really confused on how costs get calculated, but this makes it clear now.