The work behind the running application.

Every Ankra Playground size costs less than the Hetzner server of the same shape, with Kubernetes, TLS and storage included. The math, and the fine print.

We turned a production OpenSearch stack into a reusable Ankra stack profile. The parameterisation exercise turned into an audit, and it found credentials that went nowhere, a security config that was never applied, and a retention policy pointed at every index in the cluster.

What 10 heavy Claude developers cost, why one 8xB200 server cannot run Kimi K3, and the UpCloud cluster that could.

A compose stack on a VM, a Kubernetes cluster on Ankra, and the databases in between. ankra migrate up plans, converts, deploys, dumps and restores, then does it once more for the cutover. Here is what one command has to promise, and how the CLI keeps that promise.

We gave an AI agent one paragraph: build an animated deep sea facts page, wire it to our private GPU, ship it. It came back with a live URL. Here is how.

From an empty UpCloud account to a Kubernetes cluster running a team of AI agents, each with its own personality, chat channel and keys, all building through GitOps.

We ran the same k6, iperf3 and fio suite on EKS, GKE and Hetzner and lined the results up against UpCloud, OVH and DigitalOcean. The cheapest node in the test won more than it had any right to.

Two identical cloud VMs measured 2.8x apart on CPU; one failed etcd's fsync bar. A 15-minute in-cluster benchmark suite catches this before production does.

UpCloud vs DigitalOcean vs OVH on identical Kubernetes clusters. UpCloud led on storage and network, OVH on latency, DigitalOcean had the slowest tails.

ingress-nginx is retired and its successor was cancelled before it shipped. A migration path to Gateway API that does not require a big-bang weekend.