Deployment strategies

Source: Dev.to

Rolling Update

What it is
Kubernetes gradually replaces old Pods with new Pods, providing zero downtime when configured correctly.

How it works

• Some old Pods are terminated.
• New Pods are created step by step.
• Traffic is shared during the transition.

Key settings

strategy:
  type: RollingUpdate
  rollingUpdate:
    maxUnavailable: 1
    maxSurge: 1

Pros

• Zero downtime (default behavior).
• Simple and production‑ready.

Cons

• Old and new versions run together, which can complicate database/schema changes.

When to use

• Most web apps, APIs, and microservices.

---

Recreate

What it is
All old Pods are stopped, then new Pods are created.

Config

strategy:
  type: Recreate

Pros

• Very simple.
• No version mixing.

Cons

• ❌ Downtime – users see a service interruption.

When to use

• Single‑instance apps.
• Development / test environments.
• Applications that cannot run multiple versions simultaneously.

---

Blue‑Green

What it is
Two environments are maintained:

• Blue – the current version.
• Green – the new version.

Traffic switches instantly from blue to green.

How to implement

1. Deploy the new version to a separate set of Pods (green).
2. Update the Service selector (or Ingress) to point to the green Pods.

Pros

• Instant rollback.
• No downtime.
• Very safe deployment.

Cons

• Requires double the resources while both environments exist.
• Typically manual or tool‑driven.

When to use

• High‑risk releases.
• Critical production systems.

---

Canary

What it is
The new version is released to a small percentage of users, then traffic is gradually increased.

How to implement

• Deploy fewer replicas for the canary version, or
• Use traffic splitting via Ingress or a Service Mesh.

Pros

• Very low risk; bugs are detected early.

Cons

• More complex setup.
• Requires monitoring and metrics.

When to use

• Large user bases.
• Performance‑sensitive applications.

---

A/B Testing

What it is
Different users receive different versions based on headers, cookies, or regions.

Pros

• Enables business experimentation and feature comparison.

Cons

• Complex setup.
• Not a pure “deployment” strategy.

When to use

• Feature testing.
• Product experiments.

---

Strategy Comparison