How (and Why) Waze and Netflix Use Spinnaker to Breeze Through Deployments
by Jenny Medeiros
If you’re in the software industry then you’ve probably already heard of Spinnaker, the open-source, multi-cloud continuous delivery platform. It was created at Netflix from the ground up and is backed by tech giants like Google, Microsoft, Oracle, Symantec, and plenty of others.
Whether you’re already using Spinnaker or wondering if it’s time you started, you’ll be curious to know how other companies are using it to streamline their own deployments. Here are two large companies that are proud to depend on Spinnaker for their most important software process.
Let’s get straight to Spinnaker’s obvious #1 fan: Netflix. With around 100 million users and almost 4,000 deployments a day, a company like Netflix can’t afford to have a single deployment go sideways. They have to be fast, reliable, and most of all, safe.
According to their ebook on continuous delivery with Spinnaker (which you can download here), Netflix uses the platform across the board to power 95% of all their cloud-based infrastructure. Although thanks to Spinnaker’s pluggable architecture, each Netflix team is free to fashion the pipeline as they see fit.
The roll-of-the-mill kind of deployment at Netflix begins at the baking stage (their typical asset is AMI) while they run some integration tests and unit tests in parallel. Then they deploy a canary (to test the application in production using production traffic), after which there’s a manual judgement. They wait a couple of hours, maybe clean up the old server, and that’s that. This is what happens close to 4,000 times a day.
At Netflix, it’s become almost “boring” to deploy software changes because of just how easy it is. (If you want to get the full details of their deployment process, check their Senior Software Engineer’s explanation at the Armory-sponsored meetup here.)
In case you’ve only heard of Waze but don’t really know what it does, it’s a community-based traffic and navigation app where drivers give real-time info to save other drivers from road-rage.
In a blog post, various Waze engineers explained why and how they use Spinnaker to leverage their deployments in Google Compute Platform (GCP) and Amazon Web Services (AWS). They began by clarifying that Waze uses multiple cloud providers to improve the resiliency of their production systems. However, sometimes a bug in routing would make it to production undetected, so they needed the ability to quickly roll back or fix it before it caused too much trouble. This is easier said than done in a multi-cloud environment.
To their relief, Spinnaker “simplifies a lot of the complexities involved in deploying to multiple cloud platforms while keeping rollbacks easy and reliable”. This allows their developers to focus on improving Waze rather than becoming experts on the particularities of deployment.
Here’s a brief rundown of how they use Spinnaker: After the commit to git, Jenkins builds a package which is baked into a machine image on both GCP and AWS (while automated testing is also being run). The deployment then moves onto staging using blue/green strategy (aka red/black as Netflix calls it) before finally arriving to production. Thankfully, Spinnaker automatically resolves the image IDs for each cloud platform. Simple.
Breeze through your own deployments
Netflix and Waze are just two of the dozens of companies leveraging Spinnaker for smarter and easier deployments. For more info on Spinnaker, head over to their website and maybe pop into the Spinnaker Slack Channel for some friendly words with the open source community.
If you’re already using Spinnaker, drop us a Tweet on how it’s helping you and what you love most about it.
Now this invitation extends to everyone: We’d love to see you in Seattle on October 8–9 for the Spinnaker Summit. It’s going to be packed with case studies, keynotes, hands-on breakouts, and plenty of like-minded colleagues from the global open source community. If you’re into software, you really can’t miss it. Register here. (You can always transfer your ticket to a lucky friend if you change your mind!)
Jenny Medeiros is an engineer by degree turned writer by trade. She spent her first years working with Virtual Reality in South America before moving onto UX-focused Web Design and Development in Washington D.C. Now as a serial remote worker, she partners with tech-savvy companies to create content that helps people and computers understand each other better. In her spare time, she hangs with Netflix and often asks Alexa how to fold a fitted sheet.