Igor Fasano
SRE & Infrastructure Architect. Based in Portugal.
Hello. I'm Igor Fasano. I'm Brazilian, 35, and I live in Portugal, where I work for a private company in oil and gas. I have spent more than fifteen years in infrastructure.
I started with virtual machines and backup, spent a long time in the Microsoft world and then in Azure, and picked up identity and security along the way. Today I work as an SRE and infrastructure architect, and that is what this site is about.
The earlier years are not decoration on the CV. They are the reason the work goes better now. When you have actually run the hypervisor, the backup platform, the directory and the cloud subscription, you know what those systems lie about, which of their metrics mean nothing, and which single field is the one worth an alert. Monitoring a layer you have never operated tends to produce dashboards that look busy and answer nothing.
How I think about SRE
SRE is usually sold as a job for big tech companies with thousands of developers. I see it as something simpler and bigger at the same time: an operating system for the company. Every area runs better on top of it.
Security gets better when its numbers are watched all the time instead of checked once a year. Governance gets better when rules are checked by machines instead of spreadsheets. FinOps gets better when you can point at the exact machine burning money for nothing. Architecture gets better when you can see how the environment really behaves before you change it.
None of that needs a new team. It needs three things. A base: the whole environment visible in one place. A method: problems turned into numbers you can watch and fix. And now a multiplier: AI agents that use that same base to do the boring half of the work, from first triage to building the dashboard to finding what nobody uses anymore.
What I work with
Grafana, Loki, Prometheus, Tempo and OpenTelemetry, collected with Grafana Alloy and managed through Fleet Management. LogQL and PromQL. Dashboards and alert rules built from code and kept in git. On the security side, on purpose a short list: MFA, RBAC, SAML and OIDC, plus the Active Directory and Kerberos signals that sit under them. And PowerShell, because some of this still has to be a script.
On the Azure side, Azure Managed Grafana, Azure Monitor and Managed Prometheus, Log Analytics and KQL, and Azure Arc for everything that never made it to the cloud. On top of that, Azure SRE Agent with the Grafana MCP server behind it, because triage is the part I am happy to hand over first.
Why this site exists
This is my personal blog. It holds my career and whatever writing comes out of it.
The articles are about things I have actually run into, and I try to write the ones that are hard to look up. Not what the official docs say, because you can already go and read that. The part that comes after: the flag that silently returns nothing, the field that means the opposite of what it looks like, the cost nobody mentions until the bill arrives. Most of it I learned by breaking something first.
Most writing about monitoring quietly assumes cloud-native workloads, good connectivity and no legacy. Mine doesn't. The interesting problems live in the gap between what the docs say and what an environment that has to keep working on Monday will actually let you do.
It is all written to be portable. A trick that only works in one place isn't worth reading about, and it usually isn't worth doing either.
I hope you get something out of it.