You’re doing it wrong!
really; I was doing it wrong.
What are we doing wrong? Sizing of Virtualized Cloud Architecture of course! And let’s be clear, we are talking about getting the most out of your expensive Esri Licensing.
Here are two valuable documents that will change how I will architecture cloud GIS: https://aws.amazon.com/ec2/physicalcores/ AWS itemized list of actual physical cores and combines with Esri’s doc from a few years ago. https://www.esri.com/content/dam/esrisites/en-us/media/whitepaper/licensing-arcgis-enterprise-deployments-in-virtualized-and-cloud-environments.pdf We can see that you can actually use an AWS EC2 instance listed with 8 vCPU and still be legally licensing it with a 4 core ArcGIS Server license!
Example: A t3.2xlarge is a beefy server that i would have previously thought could NOT be used with a 4 core AGS license. but NOW i see that AWS only assigns 4 physical cores to that instance type. You could use that instance for Server AND Database cutting down the amount of cloud servers we need. To be fair, Esri’s document from 2018! realizes and explains this, but it’s a bit confusing in the wording. Many people would glaze over by now.
I have always recommended 4 smaller AWS servers to accommodate for most organizations 4 core limit on ArcGIS Server licensing. With this new realization (for me) a larger server would allow for more dependence on the ArcGIS Server tierof the base deployment. Instead of that box being the bottle neck (keeping it at 4 cores) this document clarifies that you can have what might be considered a massive machine with 8 processors. this will allow you to put more on that machine perhaps on cutting down of your servers.
Again. for the sake of clarity, Cloud architecture should NOT be considered an absolute cost savings over in-house hardware. It’s not a slam dunk LOTS of things need to be considered and companies spend a lot of money right sizing their infrastructure. Understanding this detail will allow for more performance for less immediate monthly cost.
Azure cloud has the same kind of explanation here.