AWS
Filed under
Cloud or On-Prem: What an SRE Checks Under the Service
The platform is your largest dependency. What an SRE verifies on a cloud or in a data center: failure domains, control planes, quotas, capacity, upkeep.
SRE in the Codebase: Timeouts, Retries, and Alerts as Code
Reliability is decided in the repository. What an SRE reviews in application code: timeouts, retries, health checks, rollouts, and burn-rate alerts as code.
SLI, SLO, SLA, Error Budget: The Numbers SRE Runs On
An SLI is what you measure, an SLO is the target, an SLA is the contract, and the error budget is the gap you may spend. The arithmetic, shown on AWS.
What an SRE Does: The Job Behind the Title
Site reliability engineering is software engineering applied to operations. What an SRE owns, what they watch, and how the role differs from DevOps.
Agent Auth on AWS: OIDC In, Access Keys Out
Agents on AWS should never hold long-lived keys. How OIDC federation, IRSA, Roles Anywhere, and AgentCore Identity turn signed proof into temporary credentials.
Agent Toolkit for AWS: The Docs Have a New Reader
AWS shipped the Agent Toolkit: an MCP server, curated skills, and IAM guardrails for coding agents. The primary reader of AWS docs is not human anymore.
FinOps: Budget Alerts with Terraform Before Your Next $2.5B AWS Bill
The July 17 Cost Explorer bug filled every feed with billion-dollar AWS bills. Ten lines of Terraform give you the budget alerts that caught it first.
Testing the Assumption Nobody Tests
AWS FIS can fail a zone and only approximate losing the control plane. How to run the game day, what to measure, and an honest scorecard for the series.
The Fix Shares Fate With the Outage
Lambda 2023, the DynamoDB DNS event, Azure's February control-plane failure: why recovery paths die with the incident, plus a lab measuring static stability.
AWS Lambda MicroVMs: Serverless Gets a Stateful, Isolated Sandbox
AWS Lambda MicroVMs bring VM-level isolation, snapshot-fast launch, and suspend-resume state to serverless. What changes, and where the Firecracker primitive fits.
Two Zones, Eighteen Hours, One Cause
Drone strikes put AWS's me-central-1 region into months of recovery and its DR advice was: restore into Europe. Plus a lab mapping AZ names to real IDs.
AWS Multi-party Approval for Organizations
Multi-party approval in AWS Organizations gates high-risk operations behind a quorum of human approvers. Here is how it works and the security it buys.
AZ Isolation Held. Your Architecture Didn't.
A cooling failure killed racks in one us-east-1 zone and a quorum died with them. What the blast radius promise covers, plus a read-only detector for it.