Google Cloud | GCE Reliability Engineering using Terraform

See where Google Compute Engine sits in infrastructure. Understand how cloud providers handle hardware and virtualization; you manage OS, runtime, and data, with PaaS examples like GKE and App Engine.

Apply GCE best practices by selecting zones near customers, distributing across zones for high availability, choosing machine types and GPUs, using labels, and avoiding external IP exposure in production.

Explore the ten stages of maturity for running apps in Google Compute Engine, from manual deployments to production-grade reliability using Terraform, managed instance groups, and cloud load balancers.

Explore three demo apps for GCE reliability engineering using Terraform: a static html page, a static html page with JavaScript, and Grafana dashboards for observability.

Set up a GCE jump box with a startup script that pre-installs git, kubectl, Terraform, and Packer, configures aliases, and enables credentials via gsutil for Terraform work.

Demonstrates end-to-end reliability engineering by provisioning three VMs in the GCP console, installing Apache on CentOS and Debian, and Grafana on Debian, configuring a data source and dashboards for observability.

Create three Debian VMs in Google Compute Engine with web server tags and labels, copying settings across instances. Verify connectivity using curl and ping within and outside the VPC.

Install grafana on a debian vm via ssh, update packages, install grafana, start the service, and access grafana on port 3000 with admin/admin credentials.

Configure static websites on GCP VMs by updating index.html with the actual IP and hostname via a bash script, restart Apache, verify with curl, and configure Grafana.

Configure Grafana by adding Google Cloud monitoring as a data source, authenticate with a Google JWT key, and import GCE dashboards to visualize VM metrics.

Complete stage one by manually creating three VMs, installing a web server, configuring websites, and setting up Grafana dashboards to view charts; prepare for stage two.

Advance to stage two by enabling logs on Google Compute Engine with the Google Fluentd agent installed on all three VMs to feed Grafana dashboards.

Showcases validating website logs from GCE VMs using Cloud Logging and Grafana after a load test, confirming log volume surge captured by Fluentd and dashboards.

Automate Google Cloud deployments with bash scripts and gcloud commands, removing console actions, to launch three servers with static HTML and Grafana, then validate observability metrics.

Set up variables and zones to create multiple VMs with Debian and CentOS images using startup scripts, then test with curl and ping before automating via bash scripts and Grafana.

Conclude stage four by deploying three servers with startup scripts and validating three apps, then introduce packer as the key tool for production workloads in gce and upcoming stage five.

Learn how Packer builds identical machine images for multiple platforms from a single configuration. Configure three images for Debian and CentOS, enabling fast, portable, stable deployment and quick testing.

Separate installation steps into a packer-built image and use startup scripts to manage boot time tasks like updating index.html and starting apache, grafana, and fluentd on centos.

Build custom GCE images with packer by taking a source image, installing httpd and fluentd, and baking new static html centos and grafana images for gcloud deployment.

Create three vms from packer-built centos and debian images using gcloud, with startup scripts to serve static html pages. Validate the deployment through curl checks, browser access, and Grafana monitoring.

Diagnose why the Apache index.html doesn't update by fixing the startup script and script name, then recreate the Debian VM and test with ping and curl.

Introduce Terraform as the next step, replacing bash scripts with infrastructure as code to reproduce Packer-built images and production-like GKE workloads.

Explore Terraform as a cloud-agnostic infrastructure as code tool for Google Cloud reliability engineering, using modules and reference docs to provision resources now and in future sessions.

Explore the Terraform reference and Google Cloud registry to reuse modules for compute resources, and master core commands init, plan, apply, and destroy while managing a state file in GCS.

Demonstrates provisioning three GCP VMs, including Grafana and static websites, with Terraform using modules, provider and backend configurations, and packaging images with Packer, including startup scripts and outputs.

Provision infrastructure with Terraform by running init, plan, and apply to create three resources; inspect state and outputs, and validate services via curl and Grafana.

Implement highly available architecture by using instance templates with a pre-configured packer image and labels to replicate VMs, ensuring immutability and reducing single points of failure.

The conclusion highlights the value of demo instance templates, how to create instances with templates, and previews building managed instance groups using reliability engineering principles for highly available applications.