The Kubernetes Developer Experience?

Kubernetes won the container wars and continues to grow in use across many industries. But how did something that was about Cloud-native Applications gain traction without a developer experience?

SHOW: 595

CLOUD NEWS OF THE WEEK - http://bit.ly/cloudcast-cnotw

CHECK OUT OUR NEW PODCAST - "CLOUDCAST BASICS"

SHOW SPONSORS:

• Teleport is the easiest, most secure way to access all your infrastructure 
• Get started with Teleport 
• CloudZero - Cloud Cost Intelligence for Engineering Teams
• Datadog Kubernetes Solution: Maximum Visibility into Container Environments
• Start monitoring the health and performance of your container environment with a free 14 day Datadog trial. Listeners of The Cloudcast will also receive a free Datadog T-shirt.

SHOW NOTES:

• Kubernetes - The Documentary - Part 1
• Kubernetes - The Documentary - Part 2
• Software Defined Talk - Eps.344 - Kubernetes Documentary

HOW DID KUBERNETES WIN WHEN IT STARTED FROM BEHIND?

Listening to this week's SDT show, and remembering listening to SDT years ago, @cote comments about why Kubernetes "won" were always interesting. In essence it was late to market, was lacking in features vs. competitors (Mesos, Swarm, CF), and had a terrible user-experience...so how did it "win"? It all seems ass-backwards. 

HOW HAS KUBERNETES CONTINUED TO WIN, WITHOUT A DEVELOPER EXPERIENCE?

1. Mesos, CF and Swarm were all single-vendor dominated projects, and many companies had concerns about another generation of vendor lock-in. This point is reasonably valid, but the companies that were using Mesos, CF and Swarm did all seem to love that technology.
2. Mesos was primarily focused on big data workloads. For each new application-type, you needed to write (or use) another application-specific framework. So it was good at its niche, but couldn't easily be used for other types of apps. [Kubernetes eventually copied this model with CRDs].
3. Swarm was the easiest to use, but it wasn't very good technology and didn't scale. So it got pigeon-holed for smaller projects.
4. CF focused on Java/SpringBoot, which is a big Enterprise opportunity. but CF was super complicated to set up. And CF never really embraced containers, so companies were weary of if they were missing this big trend (Docker).
5. Kubernetes comes along and  becomes the good-enough platform. It's not dominated by a single vendor. It natively supports Docker, it has some built-in usage patterns so it's easier than Mesos to add apps, it scales better than Swarm, and it can support Java/Spring or even legacy Java (lift-and-shift). And as Joe Beda says, you could use it natively or you could build some PaaS-y like features on top of it.

FEEDBACK?

• Email: show at the cloudcast dot net
• Twitter: @thecloudcastnet