Cloud, Kubernetes, and platform engineering.
Senior-engineer field notes on cloud, Kubernetes, AWS, Terraform, and platform engineering by Ercan Ermis.
All posts
When the Cloud Sneezes, the World Catches a Cold - Lessons from the us-east-1 Meltdown
The us-east-1 outage took down Perplexity, Signal and even the status pages meant to report it. What a single-region dependency costs, and what to change now.
AWS Monthly (Sep '25): Vega OS & eBPF Observability
In a surprise move, AWS released Vega OS in September. Vega is a specialized, Linux-based OS optimized for the edge and high-performance UI...
AWS Monthly (Aug '25): Big Data, Zero Effort
August was all about "Data Gravity." Amazon Aurora storage limits jumped to 256 TiB, which effectively ends the "sharding" conversation for...
AWS Monthly (July '25): Kubernetes at the Edge of Sanity
July was a landmark month for the EKS (Elastic Kubernetes Service) crowd. AWS announced that EKS now supports 100,000-node clusters.
AWS Monthly (May '25): The Death of the War Room
May brought CloudWatch Investigations, which have fundamentally changed someone's on-call rotation. Instead of manual log correlation, ...
Automating AWS CloudWatch Log Group Tagging with Python and Boto3
A Boto3 script that finds untagged CloudWatch log groups and applies your tag set. The IAM permissions it needs, and why untagged groups distort cost reports.
Automating AWS ECR Tagging with Python and Boto3
A Boto3 script that paginates every ECR repository and tags only the ones with no tags. The three IAM permissions it needs, and why untagged repos break costs.
Automating ECR Image Cleanup with Bash
A bash script using the AWS CLI and jq to delete ECR images matching a 9.x.x tag pattern, with dry-run and apply modes so you read the list before it removes.
Update ECR Repositories with Bash Script
A short bash loop that sets image tag mutability to MUTABLE on every ECR repository in the account, and the trade-off you accept when tags become overwritable.
Why Automated Tests Are Essential in Your CI/CD Pipeline and Development Flow
Why tests belong in the pipeline and not just on your laptop: regressions caught on every push, and a GitHub Actions job running Playwright on a throwaway EC2.
More from Ercan
Two more sites, same author, different ground.
AI, LLMs, agents, applied ML.
Field notes on AI workloads. Bedrock cost analysis, agent patterns, vector storage trade-offs, production failure modes.
Visit ercan.ai →The hub. About, consulting, contact.
Personal hub for both writing tracks. Who I am, how the consulting works, how to reach me.
Visit ercanermis.com →