Link: https://clausthorsen.wordpress.com/2015/03/29/setting-up-the-system-landscape/

From Claus Thorsen

With a system landscape, being urgently needed (see previous post), we want the construction to be of low complexity and low risk. How do we go about this?

I will claim, that one trained person with a focused effort can map a medium sized enterprise in not years, but in just one or two month! Not only that – if you cannot set aside a month, an alternative way is to map the landscape organically while e.g. supporting projects that need to understand their context. With the one-year approach, your company will not even realize the cost – it is close to zero!

Here is my approach:

1. Identify which data you need

Step one – choose which data is needed. This obviously depends on the questions you need to reply, but here is the really tricky part – you do not know! To be of general value you need to be able to reply the unforeseen. This fact may mislead you to start registering a vast number of details! Mislead because you then increase the initial scope and thereby later maintenance.

I chose to go for the following data:

• Purpose of the system
• Functionality delivered
• Data produced and consumed
• In- and outgoing data streams – which data and from/to where
• And an number of administrative data
  • Owner
  • Data sensitivity
  • Link to documentation
  • Technologies
    • System
    • Integrations
    • Infrastructure
  • Future plans

If you can easily include further data this would be excellent – e.g. system category or number of users. But make sure that it is not an excessive task to maintain the data. Also make sure not compete with e.g. the CMDB – be happy if the data are already registered. If possible (and relevant) link to existing data.

2. Identify sources knowledge

Identify who has bits and pieces of knowledge that you intend to assemble. E.g. Operation may have a list of systems; Super Users know a lot about a one or a few systems (most likely much more than you need); Your business partners know what the solution is used for. If you are fortunate, you may end up with even less than 10 persons that can help you identify and describe 80% or more!

The logical approach also might be to investigate of loads of documents. Not according to my experience! At first this takes sh.. loads of time, but much worse – the system may be used so much different from the expectation of the vendor. I have seen e.g. a project management tool used to plan and set up new stores – reading documentation would have mislead me, but a short business user interview provided all the data needed – it did not even take 30 minute, whereas tracking down just the documentation would have taken days!

3. Determine how to map (layout and tools)

A treacherous step! Take care not to let this step kill you effort.

If you in search for the right tool start out investigation, acquisition, configuration, … of the right tool, you are out of business before you even register the first system! On the other hand, a good clickable presentation tool with a structured database is surely the optimal way to go!

A very simple way to go would be to use your presentation tool at hand (e.g. Powerpoint) to register each system as shown below:

This is a very simple and efficient way to get started. The big issue is that you, to analyze the data, also need to register the same data in a structure format (database or just a spreadsheet). Unless you have a tool to help you, each registration is going to be double work.

Beside the system registration and the registry, I prefer to have a visual overview of the full system landscape. Two very different approaches are a complete registry and a system-by-system registry:

 

Each method has its peaks.

4. Go build!

Finally for the main work. In contrast to what you might expect this is actually very forward. After a few registrations, you will be able to map most systems in less than 30 minutes! 10 minutes to explain the idea, 15 minutes on the registration together with your interviewee, and finally 5 minutes to query what other systems are within the interviewee’s knowledge and appoint relevant new interviewees. 30 minutes per system!

If you track down a knowledgeable stakeholder just 3 hours of interview provides you a pretty good map of 10 systems.

That is it – the 4-step SOLID EA landscape mapping. Low risk, low complexity, and SOLID EA!

Attaining value from the landscape

All you got to do now is to harvest the value. If you look at the questions from the previous post you can answer a surprising number.

• What is the complexity of our system landscape?
• Who are the business owners?
• What are the consequences if a system is removed?

You can answer these questions straight away!

When you get questions that goes beyond the data you registered, you know exactly whom to approach.

That is it – your first true killer SOLID EA artifact!