In order to calculate and meet SLAs, it's important to consider which processes and applications are included in business use cases, especially in regard to synchronous communication. The performance of applications that synchronously call external systems directly depend on the performance of these calls. As mentioned before, distributed transactions should be avoided.

As per its nature, SLAs can only be met if all applications perform and work well together. Every application affects the SLAs of dependent systems. This not only concerns the slowest application in a system but all participating services.

For example, meeting an uptime of 99.995% per definition is not possible if it includes synchronous calls to two applications with each of them guaranteeing 99.995%. The resulting SLA is 99.99%, the values of each participating system multiplied.

The same is true for guaranteed response times. Every involved system slows down the overall response, resulting in a total response time that is the sum of all SLA times.