Once a system is configured and stable, the goal is to keep it that way, and plan for future growth. This means monitoring of uptime, and graphing of traffic, system utilization, and other factors to aid with capacity planning.
I've worked with open source system uptime monitoring products such as 'Spong', 'HobbitMon', and 'Big Brother', as well as commercial monitoring tools.
When I first started looking at graphing router traffic and terminal server utilization there were only three choices- buy an expensive commercial package that will never quite do what you want (such as HP Openview), roll your own (which is what we ended up doing at Motorola) or use a package built on GNUPlot, such as Iain Lea's 'Router Stats'.
Since then several good free and commercial products have become available, in the area of free products I've worked with Multi Router Traffic Grapher by Tobias Oetiker, and more recently I've been deploying Cricket.