Review recommended Rails CPUs for reference architectures

Through working with a customer recently it was highlighted that the rails nodes were performing very well in terms of CPU usage when running our performance tests, so much so that our current recommendations were considered to be to high by them.

Our current recommendations are based off of a lot of work from the support team working with another customer last year where they crafted out the initial form of the 10k environment.

Testing today on all of our current environments shows the rails nodes typically taking up notably less than 50% except for 2 tests that are known problem endpoints causing CPU usage to max out but these will always tax the CPU until they're fixed.

Task then is to review our current Rails CPU node recommendations and either justify their values or potentially lower them if the numbers back this.

Few factors to consider here though:

• We want to pin them to a standard GCP machine size for consistency
• Our performance tests aren't fully representative of the application (yet). They're also synthetic load tests at calculated max throughputs and as always with the nature of performance testing shouldn't be taken alone when calculating specs - It's always recommended to add in decent headroom.
• It's crucial we're able to confirm this with real world data before proceeding.