Category: platform engineering

Showing 13 posts.

K3s vs K8s vs RKE2 vs Talos: Compared by What Each Assumes About Storage

Chris Engelbert

Sep 3rd, 2026 | 12 min read

K3s vs K8s vs RKE2 vs Talos: Compared by What Each Assumes About Storage

Every Kubernetes distribution comparison scores the same things: install time, footprint, upgrade model, air-gap story. None of them score the one default that decides whether the cluster can run production VMs, and it is set before you deploy a single workload. Here is what k3s, upstream Kubernetes, RKE2, and Talos each assume about storage out of the box, and what that assumption costs on Day 2.

Why CSI Alone Can't Absorb CRUD Churn at Scale: Provisioning Latency and Day 2 Operations

Rob Pankow

Jul 27th, 2026 | 9 min read

Why CSI Alone Can't Absorb CRUD Churn at Scale: Provisioning Latency and Day 2 Operations

Every CSI provision, snapshot, and clone request is two calls, not one: a Kubernetes API object, then a separate round trip to an external array controller. At scale, that second hop becomes the bottleneck. Here is why CRUD churn breaks it, and what closes the Day 2 gap CSI leaves open.

How to Choose Persistent Storage for Kubernetes: A Decision Framework

Rob Pankow

Jun 8th, 2026 | 11 min read

How to Choose Persistent Storage for Kubernetes: A Decision Framework

Choosing Kubernetes persistent storage is a multi-dimensional decision: performance, data protection, cost, operations, and compliance all point in different directions. This framework gives platform engineers a structured way to evaluate storage options against the workloads they actually run.

We Break Our Storage So You Never Have To

Raunak Jalan

Jun 5th, 2026 | 13 min read

We Break Our Storage So You Never Have To

Simplyblock runs 100+ hours of automated chaos engineering before every release: real NVMe hardware, real FIO workloads, four failure types injected under live load. This is what we test, why it is necessary, and what it means for your infrastructure.