Deployment Strategies

Big Bang Deployment

Deploy all changes at once to all servers.

v1.0 → v2.0 (all users, instantly)

Pros: Simple Cons: High risk, difficult rollback, full impact on users

Rolling Deployment

Gradually replace instances with new version.

Instance 1: v1.0 → v2.0 (users switch)
Wait for health checks...
Instance 2: v1.0 → v2.0
Instance 3: v1.0 → v2.0

Pros: Zero downtime, gradual rollout Cons: Complex orchestration, two versions running temporarily

Blue-Green Deployment

Maintain two identical environments.

Blue (v1.0) ← Current traffic
Green (v2.0) ← Testing

Verify v2.0 works → Switch traffic to Green

Pros: Zero downtime, instant rollback Cons: Double infrastructure cost

Canary Deployment

Route small percentage of traffic to new version.

v2.0 → 5% of users (Monday)
     → 25% of users (Tuesday)
     → 50% of users (Wednesday)
     → 100% of users (Thursday)

Pros: Low risk, quick rollback, real-world testing Cons: Complex monitoring, gradual rollout time

Feature Flags

Toggle features without deployment.

if (featureFlags.newCheckout === true) {
  // new checkout flow
} else {
  // old checkout flow
}

Pros: Immediate rollback, independent feature control Cons: Code complexity, flag management overhead

Database Migrations

Careful approach to schema changes:

1. Deploy backward-compatible database changes
2. Deploy application changes
3. Cleanup old database structures

Rollback Strategies

Database

• Maintain rollback scripts
• Backup before migrations
• Test rollback procedures

Application

• Previous version still available
• Traffic routed back quickly
• Verify rollback completeness

Choosing a Strategy

• Small changes: Rolling deployment
• Complex changes: Blue-green deployment
• Risk-averse: Canary deployment
• Quick iterations: Feature flags

Best Practices

• Automate deployments
• Monitor before, during, and after
• Have runbooks for rollbacks
• Test deployment procedures
• Communicate with stakeholders
• Practice deployment procedures