Hi, I am looking for best practice suggestions here. We are dealing with a customer with a large IT landscape with large number of applications running mostly legacy technologies (JBoss 5, 6, OpenJDK 1.6 to Open JDK 1.8, Tomcat with JDK 8, Azul Zing JVM, small chunk of Pivotal implementation). Many of these apps are tightly integrated sharing data through web services, direct DB access across applications etc. We also have large number of external 3rd party integrations.
From a user perspective these are eco systems within IT landscape. So they see it as handful of eco systems. From IT maintenance and production support perspective, we need to know the application to application interactions, which application in the eco system has a problem (so we can involve right SMEs) which 3rd party service is failing and so on.
Given this context how should we define applications in Dynatrace? Should we create
Option 1: application per actual application (that is each codebase maintained as an application/service with bunch of servers running one or more instances of that app) or
Option 2: custom application per eco system, and somehow show the individual applications inside the eco system as a service end points?
Ultimate goal is to be able to track the availability of each app instance/ app as a whole / eco system as a whole / IT landscape as a whole.
Next challenge is the entire landscape has multiple regions (UAT, Integ etc) and we need to track each environment separately as we have different SLAs around detection time and resolve time (MTTD and MTTR). I am thinking management zones can help to define these?
In general there is no simple answer for this question. Everything depends of client. In our case, we used the most option to auto detect applications and then we've combined them via host name / uri into logical blocks that can be called somehow. This works for us. It's important to has proper granulation of applications because each of it is baselined separately. Monitoring team in our case is using application overview dashboard most of the time.
Remember that in Dynatrace, an "Application" is a construct that is generally tied to user actions coming from either a Web Application via Browser or from a Native Mobile Application. What you are describing can probably be organized via a combination of tags and Management Zones, which are highly flexible and can accommodate entities (such as Services) being mapped to a single Management Zone or multiple. And tags are just labels that can be added at will, so each entity can be tagged with potentially many tags,