måndag 10 juni 2013

Create an IT-service in 90 minutes

The task of creating IT-services could be daunting and it seems as many companies and IT departments are struggling with it. I like to simplify things to the point it gets ridiculous and start from there. If this mindset is applied to service and service modeling a couple of questions come in my mind. I think they are fundamental when defineing a Service in its simplest form. One important thing with these questions is that the answers can not be IT guessing. The answers must come from the IT consumer. No interpretation is allowed and I will get back to why I think so.

- WHAT is the consumer doing (in its context and with what applications)?
- WHY is the consumer doing it (what is the end result of the task)?
- What would happen if the consumer COULDN'T do the task at that moment (is it critical or could it be done later or in batch)?

So the header said "in 90 minutes" and that would imply that there is a recipe for achieving an IT-service in 90 minutes. Could it be done? well as I said, I like to simplify things so if you only allow 90 minutes to achieve this that is what you will get. On the other hand if you allow 4 weeks for it I am not so convinced that your result will be a hundred times better so keep it ridiculously short. 

What you will get is a clear understanding of what Infrastructure components are in direct relation to what the business actually do and what the effect is when they are not available. As I see it, an IT-service in its simplest form enabling IT to communicate with the business about services they can identify with and the ability to understand how it is delivered and what IT is contributing to.

Plan the 90 minutes in 2 steps. The first step is to schedule a meeting with a IT consumer that works in the area where you want to start your service modeling. The second step is to schedule a meeting with the technical responsible, probably more than one, of the applications that the consumer is mentioning during the first meeting.

Step one, meeting 30 minutes with the business. The person or persons that are invited should be very familiar with the business process and it is important that they are executing a part of their process. Try to identify a critical step in their process and ask them the three questions. you could end it right here with the answers or continue to ask them the three questions for each step of the business process that they can cover. Keep it simple and scoped instead of trying to cover too much.

Answers from first question: The answers of the first question is what I like to call service functions. Each consumer step is a potential service function. Sometimes multiple answers might be good to add to the same service function depending on the answer related to the depth in the business process. The key here is that the consumer can fully identify herself with the name and description of all the service functions. This still has nothing to do with IT so do not change it in any way to correspond more or less to IT terms, even if the consumer uses a "system" name in it, do not change it. Additionally, every time the consumer mentions a system or application, take a note of it. This will be the base for your second meeting. From a IT perspective the naming of the service functions are completely irrelevant. It could be called anything as long as the business and consumer understands what It is when using it during a follow-up or saying that it is not available for any reason.

Answers from second question: The answer of the second question is used to name the service. The logic here is that all answers from question one that is related to the same answer from question two is same as multiple service functions related to a service. If you merge two services to one then you have to relate all service functions to the new service as well. Important here to keep the business language and the correlation to the business result.

Answers from third question: The answer from the third question is the criticality of each service function. Yes that is correct, the criticality in this case is specified on service function and not the Service. This way they can easily be aggregated to the service when needed. Make it easy and classify the service requests as critical or not. The point here is that critical is a process stop for the business. A service function or business step that can be performed at another point in time is not critical.

Summary of step one, first meeting. Visually now we should have something like this.
We have one (or multiple if you had the time for it) service and at least one service function, hopefully more, related to each service. The service should have a name that completely corresponds to a result that the business is dependent on and should be identifiable by anybody relevant in the businnes. The description of the service must cover all the service functions that are related to the service but no more. The service functions are the content of the service so the service does not enable more than specified through the service requests.

Step two, meeting 60 minutes with the technical responsible (systems and applications mentioned during meeting one). The persons that are invited should be technically responsible or technically involved to the degree that they are able to answer on the technical impact of a service function in their respectively systems. The questions for each service function to the system representatives are:

- Specify any infrastructural equipment related to perform a specific service function. If you need to keep this easy, concentrate on applications installed on servers and the server itself.
- Is the infrastructure equipment critical to perform the service function or not. Can the consumer carry out the service function when the infrastructural component is not available (it might have redundancy or a queuing functionality that does not require full availability) or not.

Summary of step two, second meeting. Now we should have something like this.

Each service function has critical or non critical infrastructure components related. The infrastructure criticality only shows if the equipment is critical for the service function to work. It does not say if the service function is critical to the business. The relationship from the service function to the service tells if the service function is critical to the business. 

If you would use this visual structure when ever planing a change or managing an incident it is really powerful in impact analysis. It tells us clearly what service function a infrastructure component will affect and the meaning of it for the business/consumer. The same goes when a new demand of functionality is requested. It is easy to define if it is an add to an existing service function or a additional service function that needs to be defined and if the existing infrastructures is reusable or need additions as well.

After this first simplified step it is just the rest left :)

3 kommentarer:

  1. Hello Mika
    thanks for your thoughts. What I would like to understand further is the non-critical infra components to an IT service. If they are not critical, that is, they are not needed, why are they there at all? For me, any component in the well designed infrastructure is critical.
    Keep on blogging - keeps my mind in gear to read your views!

  2. What i mean with non critical is infrastructure that contribute and are necessary for the service but does not impact on service availability or quality. It can be redundant infrastructure solutions where you have a load balancing functionality as an example. Thanks for commenting.

  3. Nice blog! You are a great communicator Mika, your articles are very useful. Thanks and keep on blogging!