We have HA configuration architecture with two ArcGIS server (ArcGIS Site) and two portal machines. In the setup ArcGIS Data store is installed on these two ArcGIS server machines . Although it's working but all documentation suggest two different machines for ArcGIS Data store. So my question is
1. What is recommended setup
2. What are the disadvantages of current setup
It is recommended to have a primary and standby ArcGIS Datastore as you will want to data data redundancy. Without a standby data store machine you would have no redundancy for your hosted feature layer data. If the machine with ArcGIS data store fails to be reached, then you will be unable to access your data.
While it is certainly recommended to dedicate a machine for each component of ArcGIS Enterprise software, that isn't always necessary based on the usage. The major disadvantage I could imagine is that you are doubling-up on the resource demands during high loads on the hosted layers. That would cause CPU/memory usage to spike due to both web requests through the ArcGIS Server site, but also the queries that are being performed on the underlying data connection (ArcGIS Data Store in this case). The general rule of thumb is to size the machine according to our minimum system requirements for each component (which is additive), but also adjust upwards according to general usage and peak demands. With that being said, I've deployed many single-machine base enterprise deployments that are perfectly well-suited for testing and development, while production demands may be increased depending on the web-clients' usage.