Feature Flagging for Back-End & Microservices
Feature Flagging for Back-End & Microservices
I sat down with LaunchDarkly Lead Software Engineer Patrick Kaeding to get his take on feature flags for the back-end and how feature flagging works with microservices.
Join the DZone community and get the full member experience.Join For Free
The need for DevOps innovation has never been greater. Get the results from over 100 business value assessments in this whitepaper, Digital Darwinism: Driving Digital Transformation, to see the positive impact of DevOps first hand.
How does feature flagging help in back-end projects?
It helps, in general, to be able to rollout to a small set of canary users and see if things go wrong. You can gather feedback from performance metrics or error logs. You can give a smaller customer access to the new feature or even a group of customers who know they are testing out something bleeding edge. Then you can make changes and react to the feedback and then roll it out to all users or roll it back. This helps mitigate risk and lets everyone sleep better at night.
What was it like before you used feature flags?
You would have to make a change—and test it, of course, that part hasn’t changed—but then you would release it out to all of your users. You’d be watching everything to make sure nothing went wrong—and it would be a lot more stressful. You might have to time your deploy for a low traffic time, which meant staying up late on the weekend to be able to do the deploy at that time. And if something went wrong, you’d have to rollback the deploy. And if there’s any sort of data model change, sometimes rolling back to the old code is easier said than done. You might have to make sure you have a reverse migration in place and that adds a whole lot of complexity. Of course with feature flagging, you still need to make sure both code paths still work in the data model. That’s still something you have to think about.
Tell me about how microservices comes into the picture.
Microservices are great in how they help you split up your application into more manageable pieces. But sometimes, if you have a feature flag, it affects some big cross-cutting concern that would affect multiple microservices. It’s important that you keep those feature flags values in sync. And that can be tricky with some of the open source options out there —especially if these microservices are written in different languages on different platforms. And being able to write different parts of your application in different languages or using different platforms is one of the huge wins of microservices! But if you were then going to feature flag across those different systems using some open source solution that is restricted to one language, you’re going to have a bad time. So something like LaunchDarkly that has the polyglot SDKs with consistent results is going to be incredibly helpful for that…to make sure your user gets the same feature flag value on one microservice as another.
Wow! That could really be a big deal for certain users.
It’s really an orthogonal change to microservices. Certainly you can use microservices just fine without LaunchDarkly. It’s really when you want to start using feature flags in microservices that you need to come up with something to manage them.
To use all these different libraries together doesn’t even seem possible?
Yeah, especially if they are different languages. If you have them all in the same language but they’re just different services, maybe you could get away with using some shared database. But that still means that you have to make a call to the database to look up a feature flag value. For low volume that might be perfectly fine, but if you’re getting into high volume site you’re not going to make be able to make that make database calls and still serve the requests in a reasonable time. So, being able to evaluate that feature flag in-memory on each service means that you can do what you need to do much more quickly while still having control of the feature flag.
Do you have any stories about feature flagging in one of your back-end projects?
The Mongo migration is probably the most interesting story. Another pretty cool thing that I think is underrated is our New Relic integration so you can see how performance metrics are affected by feature flags. Measuring the impact of a new feature on KPI business metrics is one thing, but it is also critical to be able to monitor the operational impact. Does the new feature slow down the request serving? Or cause memory usage to balloon out of control? You need to be able to keep an eye on these things.
Are there other tools you like using with LaunchDarkly?
One cool one is DeployHub. So feature flagging lets you separate your deploy from your rollout, and LaunchDarkly gives you the view of what’s rolled out and how all that works. We also use a tool called DeployHub to track what’s deployed and when, and what code is on what server; it’s immensely helpful for that.
Published at DZone with permission of Andrea Echstenkamper , DZone MVB. See the original article here.
Opinions expressed by DZone contributors are their own.