Better performance, lower cost and less code complexity

In May 2009, Google hosted an internal "Design Wizardry" panel, with talks by Jeff Dean, Mike Burrows, Paul Haahr, Alfred Spector, Bill Coughran, and myself. Here is a lightly edited transcript of my talk. Some of the details have aged out, but the themes live on, now perhaps more than ever.

In this post, we are going to dive a bit deeper into CPU and share some general recommendations for specifying CPU requests and limits. We will also explore the differences between using the default policy (CFS quota) and the CPU Manager’s static policy. We are not going to consider memory resources in this post.

Scaling teams of site reliability engineers comes with many challenges. Here, explore the challenges of scaling and review a successful scaling framework.

How Netflix’s Container Platform Connects Linux Kernel Panics to Kubernetes Pods

Kubewatch is a Kubernetes watcher that publishes notifications to available collaboration hubs/notification channels. It watches the cluster for resource changes and notifies you through webhooks.

Since a little more than two months ago, Healthchecks.io has been sending transactional email (~300’000 emails per month) through its own SMTP server. Here are my notes on setting it up.

We’ve been using Kubernetes since before it was a “thing”, and as of 2023, we believe that it is still underutilized. In fact, it’s the best (and basically only real “at-scale”) solution for orchestrating Docker containers—or containers in general, after you’ve outgrown services like Heroku or Fly.io! That’s a bold claim, but it’s a belief backed up by our years of SRE experience. In this post, we’ll expand on that, and we’ll introduce a Kubernetes toolkit we already use and support for our clients, which simultaneously de-complexifies and highlights the benefits of Kubernetes.

OpenTofu version manager inspired by tfenv

Introduction to SLI, examples, counterexamples and tips

An error budget is essentially the permissible limit of risk or failure that a service can tolerate while still meeting its objectives. It is closely tied to Service Level Objectives, which define the expected level of service reliability. For instance, if an SLO dictates 99.9% uptime, the error budget allows for a 0.1% margin of error or downtime.

GitHub uses MySQL to store vast amounts of relational data. This is the story of how we seamlessly upgraded our production fleet to MySQL 8.0.

IaC is one of the key DevOps practices, and AWS CDK & Terraform are both great IaC tools to manage your AWS infrastructure. Having used both extensively, let me share my experience with the 2 IaC tools.