Lab: Elasticity and high availability with Docker Compose

What you will learn in this TP

• Vary the number of instances
• See load distribution
• Check fault tolerance

Horizontal scaling #

What you will learn in this section

• Vary the number of instances
• See load distribution
• Check fault tolerance

Context. The site must withstand a traffic spike then return to an economical size afterwards. That is elasticity: you simulate it by adding then removing instances behind a load balancer, and by causing an outage.

Prerequisites: Docker Compose  ·  Estimated time: 35 min

An nginx spreads load across replicas.

1. 1. Deploy the service and the load balancer

   Why? A load balancer in front of the service makes instances interchangeable: clients know a single address.

   A whoami service and an nginx using Docker DNS.

2. 2. Go from 1 to 5 instances

   Why? Going from 1 to 5 instances with one command shows horizontal scaling: capacity adjusts to demand without touching code.

   Increase the number of instances without downtime and observe the distribution.

3. 3. Simulate losing instances

   Why? Losing instances with no visible error demonstrates high availability: redundancy absorbs the failure.

   Stop two instances: does the service stay available?

4. 4. Scale back in

   Why? Scaling back when load drops is half of elasticity: that is what really reduces the bill.

   Scale in to 2 instances: how does that relate to pay-per-use?

Difficulty level : ●●○○○ (2/5)