Observability on Danilo Falcão da Silva

NewRelic vs Datadog in 2026: My Opinionated Choice

Tue, 26 May 2026 18:17:06 -0300

I’ll make this simple up front: I love New Relic for APM. I also love what Datadog built as an integrated operational platform.

If you only want a tribal answer, stop here: for app-centric teams with strong APM focus, New Relic still punches above its weight. For platform-heavy operations across infra, security, delivery, and incident flow, Datadog usually wins.

But the right answer depends on team shape, telemetry habits, and how much operational burden you are willing to carry.

The Kubernetes Operator's Batman Utility Belt: Day-2 Tools That Actually Earn Their Keep

Mon, 25 May 2026 14:35:00 -0300

kubectl is the Swiss Army knife. Nobody disputes this. But Swiss Army knives are terrible at most of the individual jobs they claim to do, and kubectl is no different: it can tail logs, but only one pod at a time. It can switch contexts, but with zero guardrails. It can describe resources, but in a wall of YAML that buries the thing you actually care about.

Day-2 operations — the part where the cluster is live, traffic is flowing, and someone pages you at 2 a.m. — need sharper instruments. What follows is the utility belt I’d recommend to any Kubernetes operator building their toolkit in 2026. Not everything here is new. Some of these tools have been around since 2018. The point is that they’re still maintained, still solve real problems, and still faster than the kubectl incantation you’d otherwise be typing.

SLA, SLO, SLI, and Error Budgets: A DevOps Reality Check

Sun, 24 May 2026 22:50:00 -0300

Most teams get SLAs, SLOs, and SLIs wrong. Not because the concepts are hard, but because they treat them as compliance paperwork instead of operational tools. The result is dashboards nobody trusts, targets nobody chose deliberately, and on-call rotations that burn people out chasing noise.

This post is a field guide for teams that actually run production systems and want reliability engineering to work as an engineering discipline — not a slide deck exercise.

eBPF Is Eating Kubernetes' iptables Plumbing

Wed, 20 May 2026 19:30:00 -0300

For most of Kubernetes’ life, the cluster data path has been a tower of iptables rules. Pod-to-service routing, NAT, network policy, even the way kube-proxy programs a Service IP — all of it expressed as netfilter chains evaluated linearly on every packet. It worked. It also aged badly.

In 2026, the answer the ecosystem has converged on is eBPF, and the project doing most of the convergence is Cilium. The shift is no longer aspirational: kube-proxy itself shipped an nftables mode that is expected to go GA in Kubernetes 1.33, the old IPVS backend is deprecated as of v1.35, and the major managed Kubernetes providers (EKS, GKE, AKS) all offer a Cilium-powered data plane as a first-class option. Azure CNI Powered by Cilium is GA on K8s 1.33.