MBMihir Bhavsar
All work

A multi-AZ AWS platform, built and shipped from code

A fault-tolerant WordPress platform on AWS, defined in Terraform and shipped through a Terraform, Ansible and GitHub Actions pipeline.

Measured outcomes

  • 99.9%

    uptime on a multi-AZ WordPress platform

    Auto Scaling across availability zones behind a load balancer, with RDS for the database.

  • <30 min

    to stand up a new environment

    The platform's infrastructure is defined in Terraform; plan and apply builds it, and one step replicates it.

  • 50%+

    faster feature delivery

    An end-to-end pipeline: Terraform provisions, Ansible configures, GitHub Actions runs it.

Context

At Parkar Global Technologies I designed and deployed a WordPress platform on AWS built to tolerate faults: it runs across availability zones behind a load balancer. Its infrastructure is defined in Terraform, so a new environment comes from code and can be replicated in one step.

Architecture

The WordPress platform, the pipeline that ships it, and what watches it

Pick one to light up the parts behind it.

  • CloudFront + S3
  • AWS region

    • VPC

      • Application Load Balancer
      • Auto Scaling group · EC2 across availability zones

        • EC2 instances · WordPress
      • RDS · database
  • GitHub Actions pipeline

    • Terraform · plan and apply
    • Ansible · configuration
  • CloudWatch · alarms, dashboards
  • Prometheus + Grafana · dashboards, alert rules

Connections

  • CloudFront + S3 to Application Load Balancer
  • Application Load Balancer to Auto Scaling group
  • EC2 instances to RDS
  • Terraform to AWS region
  • Ansible to AWS region
  • CloudWatch to AWS region
  • Prometheus + Grafana to AWS region

CloudFront, with S3, sits in front of an Application Load Balancer, which spreads traffic across EC2 instances in an Auto Scaling group that spans availability zones. WordPress runs on those instances, with its database in RDS.

A GitHub Actions pipeline ships changes end to end: Terraform provisions the infrastructure with plan and apply, and Ansible handles configuration. CloudWatch alarms and dashboards watch the platform, and I built Grafana dashboards and alert rules on a Prometheus stack that was already running.

What I built

My part

  • The platform's design and deployment: EC2 Auto Scaling across availability zones, the Application Load Balancer, RDS, S3 and CloudFront.
  • The Terraform that defines it, run with plan and apply, which lets the environment be replicated in one step.
  • The CI/CD pipeline: GitHub Actions running Terraform and Ansible.
  • CloudWatch alarms and dashboards, and Grafana dashboards and alert rules.

Already in place

  • The Prometheus stack that the Grafana dashboards and alert rules are built on.

Results

The figures at the top of this page measure three things: uptime on the multi-AZ platform, how long a new environment takes to stand up, and how much faster features ship. One result has no number: an environment can be replicated in one step.