The Cloudcast 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?
- 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.
- 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].
- 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.
- 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).
- 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