Maaz Zindani

Loading the index…

↑↓ move↵ openesc close4072 pages

Help for owners

Our main application has grown so big that nobody dares change it

Short answer

A large single application is risky to change when nothing tells you what a change will break. Splitting it into services is rarely the first step. First add tests and monitoring around the parts that change most, then release in small steps with a way back. Keep the system as it is until that safety net exists.

  • Written for business owners
  • Updated
  • By Maaz Zindani, Senior Software Engineer

Does this sound like your situation?

  • A small change takes weeks because everybody is afraid of side effects.
  • Every release is a big event, often at night, and something always breaks.
  • Only one or two people understand how the whole thing fits together.
  • Fixing a bug in one area causes a bug in a different area.
  • My developer says the only fix is to split it up or start again.

Why does this happen?

  1. Everything is connected to everything

    In one large application the parts usually share data and code. A change in the invoicing code can reach the customer screens because they use the same tables or the same helper. Nobody drew those links, so nobody can list them before the change.

  2. There are no automatic checks

    Without automated tests, the only way to know a change is safe is to try the whole application by hand. People stop trusting that, so changes get delayed and bundled into bigger and riskier releases.

  3. The knowledge sits in one head

    The reasons behind odd rules are rarely written down. When the person who knows leaves or is busy, every change turns into guesswork.

  4. Size is blamed for what is really missing safety

    A big application is not unsafe because it is big. It is unsafe because it has no tests, no monitoring and no quick way back. Splitting it up without those moves the same problems into several places.

What can I do about it today?

  1. Do not approve a full rewrite yetA rewrite or a split is a large decision. Ask for a written plan with the first small step and what it proves before you agree to anything big.
  2. Ask for a list of the most changed areasAsk your developer which parts of the application are changed most often and which ones break most often. Those are where safety work pays back first.
  3. Make sure you can get a copy of everythingCheck that you control the code repository, the server login and the database backups, and that a recent backup exists. Do not change anything on the live system yourself.
  4. Write down what must never breakMake a short list of the business actions that must keep working, such as taking an order, sending an invoice or logging in. Any safety work starts from that list.
  5. Ask for one change to be released in a small, reversible wayPick one low-risk change and ask how it would be released and rolled back. The answer shows how safe the process really is.

When should I bring someone in?

  • Releases keep breaking something, and nobody can say what a change will affect.
  • The developer is leaving and nobody else understands the code.
  • You are being asked to approve a split or a rewrite and cannot judge it.

How do I handle it?

  1. Read before proposingI read the code, the data model and the release process and write down where the parts depend on each other and which areas change most.
  2. Add checks where it hurtsAutomated tests and error reporting go first around the business actions you listed, so a change shows what it breaks before customers do.
  3. Release in small stepsChanges go out in small pieces, after a backup, with a tested way back. A bad release is undone rather than debated.
  4. Separate parts only when it paysIf one part really needs to be separate, for example because it changes weekly or needs to scale alone, I split that part and leave the rest alone. Each step is written up with its reason.

The price depends on what I find, so I quote in writing after a short discovery call. Tell me what is happening and I reply within 24 hours.

What do owners ask about this?

Should I split it into microservices?

Not as a first step. Services add networks, deployments and monitoring to run. If the main problem is that changes are unsafe, the cheaper first step is tests, monitoring and small releases.

Is a big single application always bad?

No. A single application is often simpler to run and to understand. It becomes a problem when changes are unsafe, which depends on the checks around it more than on its size.

Do I have to stop adding features meanwhile?

Usually not. Safety work can run next to feature work, concentrated on the areas the new features touch.

How was this page made?

I drafted this page with AI assistance. It is general guidance for business owners, not advice for your own system. Found a mistake? Tell me.

Next step

Want this off your desk?

Send me a few lines on where things stand. I reply within 24 hours.