i'm designing a complex GIS architecture that will publish more over 50 different services.
I have think that i will create a logical division and i will develop high availability architecture with 2 or 3 server for all Map Services (30 complex Map Services), 2 server for Geoprocessing services ( 20 Geoprocessing Services and 10 SOE), 2 server for Image Services (3 Image Service by satelite image that size is 150 TB)
All this services must be available for 500 total users. I have use this matematical alghoritm writed by Hussein Nasser into a public book, he is a Senior GIS Solution Architect:
this is correct? What is the best practice to calculate ram by number users and services?