My boss wants to document service-oriented application dependencies, going in both directions, to be used for risk analysis when we plan maintenance windows. For instance: if I take this SQL server down, what applications depend on it? If I take this Windows service down, what applications depend on that? If this switch goes down, what applications will we lose?
I pointed out to him that we had some of that functionality with manually created NPM groups and dependencies. But I think that NPM, APM, and NCM have enough information to do most of that automatically.
So what I'm proposing is a Dependency Walker/Checker tab. The more Orion addons the customer has, the richer the dependency could be. For instance, if they have NCM, Orion should know what applications will go down when a switch goes down because is knows what nodes are behind that switch. If they have APM, it knows what services or daemons behind those nodes depend on that switch.
So far it's just taking existing information and presenting it in a useful way. But an added layer that to could be if Orion could scan event logs/syslogs and infer dependencies from that. For instance, Exchange server A says in it's Application log that it can't start InfoStores because of a dependency on server B. Orion should be able to scrape those event logs and make note of that dependency. We should be able to drill up and down these dependencies, and export them as PDFs or Excel or whatever. There should be a box on any given node details listing other servers and applications that depend on that node to be operating.
