A Recipe for Architecture Clean-up

How to create order from disorder

At regular intervals, we architects need to clean up messy situations. Systems proliferate like weeds. Integrations become a tangle of underground and aerial roots. Databases and file systems bloat like dead pigs.

Over the years, I’ve refined a technique to clean up the overgrown garden and get a lovely scene of neat, well-tended, healthy plants, trees, and borders.

It works in most cases to go from information to analysis to optimisation.

There are four stages to it.

First, we use or create a good taxonomy.

This is paramount. Classifying things is one of the highest capabilities we have evolved. It’s a powerful tool to create and maintain order when put to good use. The better the taxonomy, the faster bewilderment resolves.

For example, I use types for business functions, systems, integrations, and information repositories.

It is delightful when I have to create a new taxonomy or extend one. I enjoy this phase the most and savour it while it lasts.

Second, we identify or discover all the objects in the field.

We find things and classify them into their types. The primary ones are user interface objects, business rules objects, data store objects, and interaction objects. Generic software (also called middleware) and infrastructure that serve primary objects can be tagged to them partially or entirely.

The discovery of hidden or obscure objects can take a while. I do it as far as possible, with all the necessary help. For example, finding all file transfers or archives may take a while. I am not always sure I’ve found every item, but I have learnt with time when to stop digging and move on.

In the third step, we put a size unit for the objects.

Size up all the objects of each type to compare them later. For example, it can be functional importance on a scale of 10, capacity, infrastructure cost, or software cost. Every type of object may have a different scale or measure.

Next, we sort each typed list of objects in descending order of size. It is a moment of high anticipation for me. It reveals if what I knew and felt was important before I started was correct or whether I’ll discover something new; either is satisfying.

The sizing and sorting allow the most rapid application of the benefits of the fourth step below. It also lets us decide where to stop the process if it gets too lengthy. It is remarkable how often the top quarter or half of the sorted objects in a list is the source of most chaos.

The fourth and last step is where we rationalise each type.

We apply the principles, standards and guidelines of the architecture practice. And common sense.

We remove redundancy and reposition functionality. We refactor bad integrations, apply standards, reuse data and so on.

Does it sound tedious? It’s not as hard as it sounds, as the classification and size sorting we did earlier channelises us. The picture simplifies in front of our eyes, cleanliness spreads, and savings appear.

This stage is not as pleasing to my heart as working on taxonomy and classification, to tell you a secret. But it sure is gratifying for my mind, being the raison d’etre of being an architect.

So there you have it, my recipe for cleaning up the IT systems landscape— Classify, Size, Sort, and Rationalise.

Footnote: Can we apply this outside IT architecture? Non-architect readers can consider it.

Visit Shashi Sastry’s website quality-thinking.com for art, music, and more.

Please consider Medium membership to clap, comment, highlight, and get full access to my stories and those of other excellent writers. It’s only $5 a month. Become a member and give it a go!