We're building a RESTFul service layer via Wordpress REST API. The RESTFul data is generated from custom posts, pages taxonomies etc of Wordpress. This data will then be stored in files as cache on the application server and served to a native app via OAuth2 authentication.
The data is cached on the server to speed up the RESTFul request/responses and reduce/remove the load on the Wordpress application. The data will also be cached on the mobile app to reduce the load on the application server. Only updated data will be sent to the app when updates take place. We're using an SQL Server database on a single instance and there will be no reads or writes to the database via the application layer from the app. This will only happen when an administrator has changed the data which will also keep the services running fast.
We want to start with 3 windows instances on a virtual machine scale set using a load balancer and looking to know what the best instance size would be to start. If we start small E.G D3_V2 could we easily bump it up to a D4_V2 or higher if needed without affecting downtime? Is the D Series the right way to go? Any feedback is greatly appreciated.