Are you considering the performance of cloud APIs

APIs have always been interesting. As a developer, you needed to figure out what you want to integrate as fast as possible. Sometimes this was a little challenging. 

I was at a customer where they had problems with the performance of some cloud applications they was using. See the video below. 

In on premise/a2a applications you able to create your own APIs or extend the existing APIs so the suit you. In cloud application, you have to rely on the APIs that the vendor has provided for you. It means that you would have consider how they perform in real life.

In most development situations, you would have a single tenant with test data. This would obviously not be the full load of your employees but just a subset of it. So you would only have 200 employees to test on. And in most cases this would work out without any problems. What happens when you go live is to get an application with 50,000 employees, and then it will take quite a bit of time to extract all of that data. It will probably time out. 

Load is also different on the productive tenant is also greater, and it should hopefully still give you room for the APIs. Under high load, the queries will take longer.

Therefore it is imperative that you learn how long time the current queries are allowed to take, what is the SLA of these APIs are and the amount of data that it is going to submit back to if you select all 50,000 employees. One solution would be to splice the data in company codes or in blocks of a thousand employees.

Have you run into any challenges when you have been working on cloud APIs where they did not perform, as you wanted them to?