API data sources

Hi SamyAbdul​,

v-saisrao-msft​' reply is the approach I would take as well.

I would not use the number of APIs alone as the deciding factor. Ten APIs can still be perfectly manageable with direct Fabric REST ingestion if Fabric is the main consumer and the APIs have straightforward authentication, pagination and rate limits.

Microsoft’s Fabric REST connector already supports pipeline Copy activity with pagination and authentication options including service principal.

I would bring APIM in when you need an API management layer rather than just an ingestion mechanism - for example centralized authentication, throttling, quotas, policy enforcement, transformations, monitoring, or multiple consumers sharing the same APIs.

For cost/performance benchmarking, I would compare:

• request volume and concurrency;
• payload size and pagination;
• API rate limits;
• retry/error rates;
• end-to-end ingestion duration;
• APIM tier/capacity required;
• number of consumers using the APIs.

Microsoft’s API Management cost guidance is useful here because APIM cost depends on the tier and capacity/request model rather than simply the number of APIs.

I would therefore start direct unless there is a clear governance or shared-platform requirement, then introduce APIM when the API layer itself needs centralized management.

AI-assisted drafting: AI was used to help structure and phrase this response. I reviewed and validated the technical content before posting.