Pressure Testing HAPI FHIR for web application??

[Keith Richards]

We are planning on using the implementation of HAPI FHIR server to maintain all patient resources for our web application. We have a local instance of HAPI and are using a Postgres db along side the fhir server.  We are segregating all app required resources from the patient resources. So think user would be stored in app db and patient would be in fhir db. We are then calling the fhir db for retrieval of any patient resources. Has anyone else implemented a similar architecture and if so was there any latency issues at scale? So one to one if we stored all our patient and application data in Postgres vs splitting the architecture this way, how much of a performance hit are taking? The arch decision is more about a data strategy vs a web app strategy.

Thanks for your help!