How to calculate overall application availability in Dynatrace? Means how to combine the availability metrics of a single application by combining all their related services and their synthetic monitor availability metric.
Solved! Go to Solution.
this depends on how you define availability of your application. It's probably not best practice to just combine each related services availability to produce application availability numbers. Outages of related services should impact differently to overall application availability. Best practice is to define application's key objectives and create synthetic tests that produce exact information about passing/not passing those objectives (f.e. ability to log in, ability to perform actions in defined time limits, etc..).
to Echo what @rastislav_danis said, its really depends on what your organization classifies or defines availability. Services, Processes and hosts all can be run on and off in tandem with load balancing for high redundancy. "availability" is more so aligned with the end user - Is is available for the end user to use, and interact with. That's where the Synthetics come into play. You can do a simple test of the site is it up? You can also do specific ones where you test a function, so your site is up, but bill payment is broken. It really depends on how your org wants to define "Availability"
While I have been a Synthetics advocate for more than 20 years now, and it seems the right way to go, it does have a lot of drawbacks in real world scenarios. I have been pushing more for SLOs with several key SLIs, of which availability is one of them. As @ChadTurner and @rastislav_danis have said, it really depends what you consider availability? If you move from availability to SLOs, as some of my clients have, you might even not be focused on how many 9s you have, but how many money you make for instance in a site... Looks like a good SLI to me 💰