Distributed Systems

Overview

This assignment introduces Go, RESTful APIs, cloud deployment, and basic performance testing.
I implemented a simple backend service using the Gin Web Framework, deployed it to AWS EC2, and evaluated its performance under load.

---

Part I: Go + Gin REST API

• Implemented a RESTful API with endpoints:
  • GET /albums → returns all albums as JSON
  • POST /albums → adds a new album from JSON request body
  • GET /albums/:id → returns a specific album by ID
• Learned about:
  • go mod init and dependency management with go get
  • Handlers, endpoints, and JSON serialization
  • Difference between localhost and 0.0.0.0

---

Part II: Deployment to AWS EC2

• Created and configured an EC2 instance (Amazon Linux 2023, t2.micro)
• Steps:
  1. Generated SSH key pair (rsa_pem.pem)
  2. Connected via SSH:

     ssh -i rsa_pem.pem ec2-user@<EC2_PUBLIC_IP>

  3. Cross-compiled Go program for Linux:

     GOOS=linux GOARCH=amd64 go build -o server main.go

  4. Uploaded binary with scp and ran on EC2
• Configured Security Group to open port 8080
• Verified deployment with:

  curl http://<EC2_PUBLIC_IP>:8080/albums

  ---

Part III: Performance Testing

• Wrote a Python script (load_test.py) to: Send continuous GET requests for 30s Collect response times Visualize results (histogram + scatter plot) Compute statistics (average, median, percentiles)
• Results: Average latency: ~65 ms Stable baseline, with a few slow outliers (>100 ms) Demonstrated long tail latency
• Insights: t2.micro has limited CPU → variability under load Tail latency grows worse with higher concurrency Network + server both contribute to delays

Demo Instructions

1. Start EC2 instance
2. SSH into instance:

   ssh -i rsa_pem.pem ec2-user@<EC2_PUBLIC_IP>

3. Run server:

   ./server

4. Test SPI:

   curl http://<EC2_PUBLIC_IP>:8080/albums

Repository Contents

• main.go → Gin REST API code
• go.mod, go.sum → Go module files
• load_test.py → Python load testing script
• README.md → this file