Since the last update I’ve written articles on a wide range of topics. The first was about creating a good DevOps resume. The DevOps job market really heated up in 2018 and LinkedIn produced a report stating that DevOps Engineers were the most recruited role of the year. Hopefully my blog will help you stand out from the crowd.
The next blog was a follow on from the Kubernetes cloud services comparison. I tested spinning up GKE and AKS clusters and created some pretty graphs. As expected AKS was significantly slower than GKE and during the 9 hour test run failed a couple of times. If you’re being forced to use Azure at work you have my condolences.
An anonymous individual from Microsoft contacted me after reading the blog results to say he wasn’t surprised. The underlying infrastructure layer runs on HyperV and it is apparently a total shit-show that nobody is able to fix. This explains the overall slowness and unreliability. Maybe it’s time for Microsoft to properly embrace Linux and get rid of the Windows layer that’s underpinning and crippling their cloud products.
While we’re on the subject of Kubernetes platforms there has been continued work on the On-Prem spreadsheet. This now includes data for OpenShift Container Platform, Rancher RKE, PKS, Docker Enterprise, Kubermatic, CDK – Canonical Distribution of Kubernetes, SUSE CaaS Platform, MetalK8s, Mesos, IBM, DaoCloud Enterprise and Platform9.
The original comparison blog is here. The comments are quite funny. The nonsense from Docker on the On-Prem blog echos the nonsense from Microsoft on the cloud blog. Microsoft really should buy Docker and form a single entity that fails technically both in the cloud and On-Prem. They can then go head to head with IBM/Redhat and race to extinction.
Finally, I wrote a blog about running Kubernetes on your laptop. Minikube and Docker Desktop came out as winners.
I’ve picked up the spreadsheet for the Kubernetes network comparison again. It’s quite big and complicated and has taken a few hours so far. Hopefully once complete it will be useful to people.
From the 44 votes on LinkedIn it seems I should also write about alternatives to Helm for packaging up applications.
There also seems to be a gap in really good training material. Or at least, there is material out there but no clear path for people to take. I’m going to look into how to organise and structure the free online material that already exists into some kind of step by step course.
Only a couple added since the last update. Magic IP Address lets you do some funky networking stuff with static IP’s. There’s also the Node Problem Detector chart that got added.
node-problem-detector, draino and cluster-autoscaler all work together to self heal applications running on broken nodes.
Almost up to 60k users since the site began a couple of months ago. The most popular blog by far is the cloud comparison blog.
People also liked the base images and security scanning blogs. I’m getting more involved with DevSecOps during my day job so hopefully I can do more on this topic in future.
It was only 30 days ago that Kubedex started to rank for the X vs Y keywords. These usually take 6 months or more for organic results. Over time I expect organic traffic to outgrow the erratic social network traffic.
That’s all for this week. Message me on Kubedex.com via the contact form or connect with me on LinkedIn if you have any feedback or ideas for cool blogs.
Project Dolos: Testing Kubernetes on Google and Azure This is part 1 of a 2 part blog where we look at Kubernetes…
Tell us about a new Kubernetes application
Never miss a thing! Sign up for our newsletter to stay updated.
Discover and share new Kubernetes applications