Last Updated on December 31, 2018
It’s holiday season and I’ve been traveling and seeing family. There has only been a single blog since the last update. If you’re interested in becoming a SRE or fancy taking the Kubernetes CKA exam then you may want to see this list of courses.
There was a brief flurry of new stable Helm charts added recently.
K8s Spot Termination Handler: Watches the EC2 metadata service and 2 minutes before termination cordons and drains the node.
OPA: Open Policy Agent lets you define a bunch of policies to secure your cluster.
Sealed Secrets: Keep your secrets in Git along with your other config and this operator will decrypt them.
Kuberhealthy: Synthetic testing that works alongside Prometheus to detect cluster problems.
To read about the others you can see all of the latest charts on Kubedex: https://lnkd.in/fX7PMSY
Google analytics is starting to show some reasonable statistics after nearly 100,000 page views. I was curious to see which topics are the most popular. Most of my traffic lands on blog articles but often people click around exploring the various categories of Helm charts.
There are sections for other clouds but nobody really seems to care about those. I predict 2019 could be a big year for EKS as the sheer number of locked-in AWS companies start to use it in anger.
When re-platforming it’s not a surprise that many people will look to modernise their CI/CD infrastructure. Jenkins hasn’t aged well and there’s really no defacto standard cloud-native CI server. Gitlab is the best of a bad bunch but is still a messy evolution of bolted on extras rather than a beautiful set of abstractions.
With tools like Prometheus, Istio, Jaeger, Elasticsearch and hundreds of other projects it feels like observability has never been better. I’d argue that the gap now is less about tooling and more about how to make use of all of the tools properly.
One thing that did surprise me from looking through the stats was just how globally distributed the audience is for Kubernetes articles.
Around a third of the traffic comes from North America followed closely by about 10% from the UK. After that there’s a long tail of visitors from almost every country in the world. Looks like 2018 was the year of global domination for Kubernetes.
There is a definite trend towards using the Kubernetes API server as a single API across all cloud infrastructure. The ability to extend Kubernetes with CRD’s and the abundance of specialised controllers (Operators) is turning Kubernetes into the control plane for all cloud resources.
I predict that the AWS Service Operator will change the way most companies provision infrastructure in 2019. Less Terraform and Ansible and more configuration through the Kubernetes API.
Similarly, 2018 was the year that everyone recommended managing databases outside of Kubernetes. 2019 will be the year that the early adopters of database operators start to get burnt before they fully mature.
Security is an area full of low hanging fruit. Containers running as root, no user name spacing and complicated seccomp and pod security policies make misconfiguration easy. This is an area where Kubernetes really needs to jump forward in 2019.
On the negative end of the spectrum I don’t see Serverless going anywhere for a while. The current set of projects don’t provide enough tangible value to use and I can’t see a path to their success. These will remain in the domain of cloud providers and will continue to be a cheap way of joining together batch operations.
We will hopefully see more progress with Kata Containers since Firecracker was released.
I’m 50/50 on ARM vs Intel for 2019. It’s not clear if the cost savings of running workloads on ARM workers would outweigh the risk of running code on a different architecture.
That’s all for this year. I have some exciting projects lined up for early 2019 and some fun new blogs in the pipeline.
Tell us about a new Kubernetes application
Never miss a thing! Sign up for our newsletter to stay updated.
Discover and learn about everything Kubernetes