Network management experts often preach the need for IT organisations to break down organisational silos so that network managers can collaborate with other groups on infrastructure management. Despite these sermons, many networking pros still struggle with this concept.
Donavan Pantke, network engineer for Appriss, a contractor that captures and configures information about inmates in correctional systems, has found a way to break down those silos using a service-based network management approach.
Pantke has been using CA Inc.'s Spectrum Infrastructure Manager for six years to manage his infrastructure in a way that hews very closely to the ITSM (IT Service Management) methodology, which is an extension of ITIL (Infrastructure Technology Information Library). The product offers a single-pane-of-glass view of his company's infrastructure. Appriss' small IT staff of 10 all use Spectrum in their different capacities as systems administrators or network managers. But they all treat the infrastructure from a service-based point of view rather than as a collection of individual components.
"We heavily use the service manager part of Spectrum to build out services of all these different components in the network," Pantke said. "So instead of having to worry about understanding the relationships between each and every component, we relate each and every component to a service. And then we track those services in uptime. We have a very highly tiered, large service structure, so if we lose a particular device in this area, I know all the upstream impacts and how the services are impacted."
Pantke said an ITSM approach allows his company to link all of its IT silos together through services. He said this allows him to figure out quickly which components are involved in services.
"One of the things we have learned as an organisation is that, in general, silos are bad," he said. "Sometimes they are a necessary evil, but in general silos enable a large section of the blame game to occur. This is a server problem. No, it's a network problem. No, it's an application problem. The buck gets passed all the way around to different people until someone gets mad enough to say, 'We have to figure this out!'"
Pantke also uses Spectrum to prioritise what he troubleshoots.
"If I have something that's impacting services, it could be impacting three customers or … 150," he said. "Even if the problem that impacts 150 is minor and the other is critical, I might want to go after this minor one first because it's got such a large enterprise impact scope to it."
Having a service-based view also allows Pantke to see problems not just with component devices but also with links.
"We do a lot of our inter-switching on fiber," he said. "We're not a big fan of copper. When you're in that mode, the No. 1 problem you have from a link standpoint is with fiber transmitters. You can get a laser that dies."
Since many of the links in Pantke's network are redundant, services aren't necessarily degraded right away by a bad link. But he can catch the failure early, before it becomes a problem. Ultimately, while Spectrum works as a network management tool, Pantke considers it an overarching solution.
"Calling Spectrum a network management tool is actually giving it less credit than it deserves," he said, "because it can manage networks, servers and, to a degree, applications."