IT system failures cause enormous financial damage. To avoid or at least shorten these, several different approaches have been developed in the past to analyse either metrics, logs, or traces. The objective of this thesis is to describe how the monitoring of IT systems can be extended by evaluating the different data types within a single software solution.
For this purpose, the term Observability is defined, as there is no generally valid definition for this at the time of writing the thesis (December 2021). Subsequently, the vendor solutions Elastic, Datadog and Splunk Observability are evaluated to determine whether they are suitable for implementing the Observability approach. To show that the approach not only works theoretically according to the vendor’s statements, a proof of concept environment is set up to demonstrate an exemplary implementation.
The evaluation shows that the Observability approach can be successfully implemented with all three vendor solutions.