IT Service Management is a collection of processes and disciplines used to manage the delivery of services covering infrastructure, applications, and business processes. When I say manage I mean ‘oversee the delivery of day to day services and ongoing change to ensure delivery of agreed services at agreed pricing and to agreed service levels’.
The most important aspects of Service Management are requirements and measurement. You need to know what you want to achieve from service management, i.e. the success criteria, and then you need to measure whether or not you’re achieving these. Of course you also need other things like the flexibility to change requirements, but if you know what you want and how to measure it you’re on the right track.
From my many years of experience in designing, selling, delivering and troubleshooting the management of outsourced services, I have arrived at a simple but powerful model that can be used to define most IT Service Management regimes.
Starting with the end users we see that what is important to them is that it just plain and simply works. And when it doesn’t work, it gets fixed. Moving up the food chain to IT management, then further still to top level management, we can arrive at a short list of what’s important and therefore what we should be measuring in any service management regime. A question-based approach appears to work. Basing a service level regime on these criteria is a good place to start. There are implications, however. Most IT service providers measure component availability in the data centre, but not application availability from the user’s perspective. Tools do exist to do this and many companies deploy them, but it is often culturally difficult to base a commercial service level contract on them. Suppliers aren’t always in control of the end-to-end infrastructure, e.g. Wide Area Network, and clients often find it difficult to let go of detailed component availability metrics. But if the critical services are available to end users for the agreed times at the agreed performance, should you really care whether all servers were available 99.99%?