Jack Marchant

Principal Software Engineer @ Deputy

Twitter | GitHub

Maintaining feature flags in a product engineering team

I have mixed feelings about feature flags. They are part of the product development workflow and you would be hard pressed to find a product engineering team that doesn’t use them. Gone are the days of either shipping and hoping the code will work first time or testing the life out of a feature so much that it delays the project.

The benefits of using feature flags certainly outweigh the bad, but it doesn’t stop teams from cursing them every time a major bug is reported or an incident occurs as a result of enabled (or disabled) feature flags.

In this post I will discuss the benefits and some drawbacks of using feature flags, to help you learn from some of the lessons I’ve personally learned, in the hopes that you can avoid the mistakes.

First, let’s understand what a feature flag does, and why it’s there:

A feature flag is (at its simplest - there are more advanced controls you can use) an on/off switch to release some new functionality to your product through the code base.

Teams can safely ship code knowing the feature can be enabled for small groups of users at a time, and released to more customers as confidence in a feature or behaviour grows.

Before feature flags, you only had one shot to ship code to production and make sure it works, which meant a longer build up to releasing code for the first time, or complicated infrastructure to support canary releases.

The main problem with feature flags is what happens when you have too many and they start conflicting with each other or you have so many different flows to test that the team spends much longer on a feature than they should.

This leads me to the first lesson:

Lesson #1

The number of feature flags you maintain will spiral out of control

Every time you create a feature flag, you’re introducing a different behaviour for your code that may only be initially released to parts of your user base, meaning you now support two different behaviours (feature flag on and feature flag off).

This is empowering for a growing product and engineering team. As the number of feature flags in use in production grows, so too does the frustration of testing all of those different cases and receiving bug reports where your first instinct is to check which feature flags are enabled or disabled.

Keeping track of feature flags means making them attributed to your team so there’s ownership of the feature flags and tracking the rollout status, including ensuring rollout continues to happen, or the flag is retired.

When feature flags have already spiralled and are out of control, the best thing to do is pause development of new features and clean up any flags that are rolled out or no longer required.

Getting feature flags back under control should be a priority, given the impact on development and testing time for related features.

This is a hard lesson to learn because there’s only one way out. Clean up the flags!

Lesson #2

Clean up feature flags regularly, make it part of the development cycle

After some time in production, feature flags can become stale and turn into technical debt, which must be paid back at some point, or risk cumulating over time and being at the point of no return. In some ways you will always have to live with feature flags, and they become part of the work you do.

The difference between a feature flag that is new and part of the feature actively being worked on, versus a flag that has lost purpose is enormous.

To avoid this dysfunctional reality, we must clean up feature flags after the feature has been rolled out or no longer needed. Sometimes this will be a minor piece of work that involves only one person making the code change and testing it and other times it can require re-testing an entire feature.

In my experience, how well you have built up your automated testing around the feature will impact whether it’s a minor change or a major one.

Recently in my team, we had built up a lot of feature flags for a variety of reasons, whether it was changing teams, forgotten features or slow rollouts and so we had to take a week or two to clean up around 10-12 feature flags in a short period of time.

We ended up doing the work for these in a short time frame, then merging and releasing the changes incrementally and over a longer period just in case anything went wrong.

This proved to be successful in the end and the team swarmed on the work to get it done.

We’re working on keeping track of feature flags more closely now to make it part of our development cycle to clean up feature flags rather than waiting for the eventual build up.

When we release a new feature, with a corresponding flag, we document it and ensure it keeps rolling out and create a ticket for the future to clean it up.

Progressing with the rollout usually means opening up the feature to more customers and this brings me to the final lesson.

Lesson #3

Always be rolling out

Feature flags should be temporary and are meant to increase the velocity of your team by allowing you to ship quickly and get real feedback from smaller groups of users in a safe way.

Flags, therefore, should be intended to be rolled out completely to your whole user base at some point. It’s normal to start with a small group and then build up incrementally to larger groups, but this should always be on a timeline.

Once you forget about it or move on to the next thing and leave the flag there, it will become stale and your team falls into the trap of having to maintain it and test both states of the flag should a change to that area of the code base be required.

There’s no one right time frame for a feature flag to exist, it will always depend on the feature and the group of customers using it to give you feedback directly or indirectly, through usage.

That’s why keeping track of the current state of feature flags in your control is important including managing the continual rollout to more users.

As you find bugs you can pause the rollout until the bugs are fixed, but if you haven’t hit any road blocks it’s critical to keep forging ahead so removing the feature flag becomes possible once it has been made available for all of your users.

Feature flags are both a blessing and a curse, it’s probably no secret to most engineering teams. What’s missing in my opinion is a framework to manage feature flags over time and throughout engineering organisations.

They help keep the product working and make it easy to rollback changes without code deployment, and give on-call engineers peace of mind when they can safely turn off a flag that has caused an incident.

If left unchecked, feature flags can slow engineering teams down to a crawl so create each feature flag with caution and a plan for its eventual removal.

Feature flags have given product teams the confidence to move fast, with a plan to rollback at the click of a button, but with great power and flexibility, comes a cost which should not be underestimated.

. . .

how does a relational database index really work

A common question in software engineering interviews is how can you speed up a slow query? In this post I want to explain one answer to this question, which is: to add an index to the table the query is performed on.

refactoring for performance

I spend most of my time thinking about performance improvements. Refactoring is tricky work, even more so when you’re unfamiliar with the feature or part of the codebase.

exploring async php

Asynchronous programming is a foundational building block for scaling web applications due to the increasing need to do more in each web request. A typical example of this is sending an email as part of a request.

maintaining feature flags in a product engineering team

I have mixed feelings about feature flags. They are part of the product development workflow and you would be hard pressed to find a product engineering team that doesn’t use them. Gone are the days of either shipping and hoping the code will work first time or testing the life out of a feature so much that it delays the project.

technical interviewing

When I first started interviewing candidates for engineering roles, I was very nervous. The process can be quite daunting as both an interviewer and interviewee. The goal for the interviewer is to assess the candidate for their technical capabilities and make a judgement on whether you think they should move to the next round (there’s always a next round). Making a judgement on someone after an hour, sometimes a bit longer, is hard and error prone.

using a dependency injection container to decouple code

Dependency Injection is the method of passing objects to another (usually during instantiation) to invert the dependency created when you use an object. A Container is often used as a collection of the objects used in your system, to achieve separation between usage and instantiation.

3 tips to help with working from home

Working from home has been thrust upon those lucky enough to still have a job. Many aren’t sure how to cope, some are trying to find ways to help them through the day. Make no mistake, this is not a normal remote working environment we find ourselves in, but nonetheless we should find ways to embrace it.

making software a three step process

One of the most useful tips that has guided much of my decision over the years has been this simple principle: three steps, executed in sequential order;

help me help you code review

Code Reviews are one of the easiest ways to help your team-mates. There are a number of benefits for both the reviewer and pull request author:

a pratical guide to test driven development

It’s been a while since I last wrote about why testing is important, but in this post I thought I would expand on that and talk about why not only unit testing is important, but how a full spectrum of automated tests can improve productivity, increase confidence pushing code and help keep users happy.

facade pattern

Design Patterns allow you to create abstractions that decouple sections of a codebase with the purpose of making a change to the code later a much easier process.

the problem with elixir umbrella apps

Umbrella apps are big projects that contain multiple mix projects. Using umbrella apps feels more like getting poked in the eye from an actual umbrella.

broken windows

Ever get the feeling that adding this "one little hack", a couple of lines of code, won't have much of an impact on the rest of the codebase? You think nothing of it and add it, convincing your team members it was the correct decision to get this new feature over the line. In theory, and generally speaking, I would kind of agree with doing it, but every hack is different so it's hard to paint them all with the same brush. If you've been doing software development for long enough you can see this kind of code coming from a mile away. It's the kind of code that can haunt your dreams if you're not careful.

lonestar elixir 2019

Last week was Lonestar ElixirConf 2019 held in Austin, Texas. The conference ran over 2 days and was the first Elixir conference I had been to.

genserver async concurrent tasks

In most cases I have found inter-process communication to be an unnecessary overhead for the work I have been doing. Although Elixir is known for this (along with Erlang), it really depends on what you’re trying to achieve and processes shouldn’t be spawned just for the fun of it. I have recently come across a scenario where I thought having a separate process be responsible for performing concurrent and asynchronous jobs would be the best way to approach the problem. In this article I will explain the problem and the solution.

best practices third party integrations

When we think about what an application does, it's typical to think of how it behaves in context of its dependencies. For example, we could say a ficticious application sync's data with a third-party CRM.

you might not need a genserver

When you're browsing your way through Elixir documentation or reading blog posts (like this one), there's no doubt you'll come across a GenServer. It is perhaps one of the most overused modules in the Elixir standard library, simply because it's a good teaching tool for abstractions around processes. It can be confusing though, to know when to reach for your friendly, neighbourhood GenServer.

offset cursor pagination

Typically in an application with a database, you might have more records than you can fit on a page or in a single result set from a query. When you or your users want to retrieve the next page of results, two common options for paginating data include:

protocols

Protocols are a way to implement polymorphism in Elixir. We can use it to apply a function to multiple object types or structured data types, which are specific to the object itself. There are two steps; defining a protocol in the form of function(s), and one or many implementations for that protocol.

exdocker

Recently, I've been writing a tonne of Elixir code, some Phoenix websites and a few other small Elixir applications. One thing that was bugging me every time I would create a new project is that I would want to add Docker to it either straight away because I knew there would be a dependency on Redis or Postgres etc, or halfway through a project and it would really slow down the speed at which I could hack something together.

working with tasks

While writing Understanding Concurrency in Elixir I started to grasp processes more than I have before. Working with them more closely has strengthened the concepts in my own mind.

understanding concurrency

Concurrency in Elixir is a big selling point for the language, but what does it really mean for the code that we write in Elixir? It all comes down to Processes. Thanks to the Erlang Virtual Machine, upon which Elixir is built, we can create process threads that aren't actual processes on your machine, but in the Erlang VM. This means that in an Elixir application we can create thousands of Erlang processes without the application skipping a beat.

composing ecto queries

Ecto is an Elixir library, which allows you to define schemas that map to database tables. It's a super light weight ORM, (Object-Relational Mapper) that allows you to define structs to represent data.

streaming datasets

We often think about Streaming as being the way we watch multimedia content such as video/audio. We press play and the content is bufferred and starts sending data over the wire. The client receiving the data will handle those packets and show the content, while at the same time requesting more data. Streaming has allowed us to consume large media content types such as tv shows or movies over the internet.

elixir queues

A Queue is a collection data structure, which uses the FIFO (First In, First Out) method. This means that when you add items to a queue, often called enqueuing, the item takes its place at the end of the queue. When you dequeue an item, we remove the item from the front of the queue.

composing plugs

Elixir is a functional language, so it’s no surprise that one of the main building blocks of the request-response cycle is the humble Plug. A Plug will take connection struct (see Plug.Conn) and return a new struct of the same type. It is this concept that allows you to join multiple plugs together, each with their own transformation on a Conn struct.

elixir supervision trees

A Supervision Tree in Elixir has quite a number of parallels to how developers using React think about a component tree. In this article I will attempt to describe parallel concepts between the two - and if you've used React and are interested in functional programming, it might prompt you to take a look at Elixir.

surviving tech debt

Technical debt is a potentially crippling disease that can take over your codebase without much warning. One day, you’re building features, the next, you struggle to untangle the mess you (or maybe your team) has created.

pattern matching elixir

Before being introduced to Elixir, a functional programming language built on top of Erlang, I had no idea what pattern matching was. Hopefully, by the end of this article you will have at least a rudimentary understanding of how awesome it is.

first impressions elixir

Elixir is a functional programming language based on Erlang. I’m told it’s very similar to Ruby, with a few tweaks and improvements to the developer experience and language syntax.

write unit tests

Unit testing can sometimes be a tricky subject no matter what language you’re writing in. There’s a few reasons for this: