Identifying and eliminating single points of failure

Initial state

• Application server running on a single VM

• DNS configured to point directly to the public IP address or DNS of the VM

• Tightly coupled setup, posing a single point of failure

Improvements

• Use of a load balancer

• Application servers

• Databases

Load balancer

Split out the reverse proxy functionality into a load balancer. It allows for multiple availability zones and offers redundancy and automatic scaling.

Application servers

Instead of a single VM, use multiple VMs for the application servers. Employ auto-scaling to scale horizontally, ensuring redundancy.

Database

Utilizes AWS Aurora serverless, a relational database service using MySQL or Postgres. Rather than running on discrete VM, it employs ACUs with CPU and memory. All ACUs are fronted by a single proxy endpoint directing queries to the next available ACU