Reliability is a system’s ability to perform its intended functions consistently over time. AWS services are often highly available or fault tolerant by default, and AWS provides tools for building reliable systems. section 1.3.5
A rolling update can preserve availability during deployment: CloudFormation launches a replacement EC2 instance, waits for it to become ready, and then terminates the old instance. Users can receive the new Etherpad version without downtime. section 15.2
Reliability depends on how deployment work is performed:
- User data can bootstrap an EC2 instance by installing Node.js and Git, downloading application code, and starting the service. section 15.2
- Deployment scripts introduce external dependencies: a repository outage can prevent source code or packages from being downloaded. section 15.2
- Building an AMI with the required application and dependencies removes those boot-time dependencies and can shorten startup. section 15.2
CloudFormation rolling update
Update the CloudFormation stack with a new Etherpad version, let the Auto Scaling group create a replacement instance, verify the application, and allow the old instance to be removed. Check cloud-init logs when bootstrap fails. section 15.2
