T-SQL Tuesday #156 Production Code

Introduction

The invitation from Tom Zíka asks “Which quality makes code production grade?“.

Reliability

Code that runs in production should be reliable. It should produce consistent results, handle concurrency and deal with error situations.

Performance & Scalability

Production code needs to be able to scale. Scalability should be considered in the context of your own applications and their expected growth. Not everyone has to deal with the same scalability challenges as Google, Facebook or Twitter.

Performance is also important. It doesn’t matter how awesome your app is if it can’t return data fast enough for end users.

Fast and efficient code can save $$$ in running costs and keep end users happy.

Security

Production code needs to be secure, following industry best practices. A security incident can cause severe reputational damage and financial impact.

Maintainability

Production code should be maintainable. Avoiding duplication and striving for simplicity.

Reviewed

Production code will often go through a code review by another developer before it’s accepted into the codebase.

Tested

Production code will go through a series of automated and human tests before it gets to production.

Accurate?

Production code doesn’t always have to produce accurate results. Accuracy can be very important in some situations. In others, you might be able to trade some accuracy for speed. e.g. Caching data, eventual consistency.
The number of active users on a website doesn’t need to be 100% accurate and can be out of date by the time it’s printed.

Documented

Production code should have comments and documentation.

Other

You might have policies for coding standards to follow. This might include things like naming conventions, code formatting and other best practices.

Does it need to be perfect?

Production code doesn’t have to be perfect. Perfect code doesn’t really exist. The code we write today will be viewed through a different lens tomorrow with the benefit of hindsight, experience and new improved ways of doing things.

The code we ship to production needs to be fit for purpose. Shipping code that works and delivers value in a timely manner is often more important than spending months trying to find the most optimal solution. There is a balance to be struck.

Architectural decisions made early in a project can have long-lasting consequences. Making good decisions early in a project can be cheaper than fixing them later.

You can’t always predict how users will use a new feature and how to optimize it. Growing pains are inevitable for successful applications. It’s important to monitor and be proactive in fixing problems.

Not production grade

So what are some examples of code that shouldn’t be in production?

  • Code that uses undocumented/unsupported features or behaviour
  • Untested code
  • Insecure code. e.g. Building dynamic SQL in a way that is prone to SQL injection.

Summary

Production code can vary in quality for a variety of reasons. The experience level of the developer writing the code, time pressures, and level of fatigue all impact code quality.

It’s not unusual for code written as a quick PoC to find its way into production. Code can also start to get messy over time as you start to shoehorn new features into an existing codebase.

If your code is delivering value to the end-user or business you can feel happy about it. Over time we all accumulate technical debt and need to allocate time to reduce it.