GCP Golang vs Node.js Performance Explorer

This project is a performance comparison experiment between Go and Node.js running on Google Cloud Run.

It consists of two functionally identical applications (one in Go, one in Node.js) that interact with Google Cloud Firestore. Both applications are instrumented with OpenTelemetry to provide deep insights into latency and execution tracing via Google Cloud Trace.

Project Structure

• go-app/: A Go web server using gorilla/mux and html/template.
• node-app/: A Node.js web server using Express and EJS.
• infra/: Terraform configuration to provision the Google Cloud environment (Cloud Run, Firestore, Artifact Registry, etc.).

Goal

The goal is to deploy both applications to Cloud Run and measure:

1. Cold Start Latency: How fast does the container spin up and serve the first request?
2. Runtime Performance: How fast is the request processing (API handling + Template rendering)?
3. Resource Usage: Memory and CPU footprint difference.

Prerequisites

To run these applications, you need a Google Cloud Project. You can set this up manually or use the provided Terraform infrastructure.

Option 1: Infrastructure as Code (Recommended)

See the Infrastructure README for instructions on how to provision the entire environment using Terraform.

Option 2: Manual Setup

1. Enable APIs: cloudtrace.googleapis.com, logging.googleapis.com, monitoring.googleapis.com.
2. IAM Roles: Ensure your deployer has Cloud Run Admin, Artifact Registry Admin (for source deploy), and Service Account User.
3. Firestore: Create a Native mode database.

Quick Start

1. Clone the repository.
2. Navigate to the desired application folder (cd go-app or cd node-app).
3. Follow the specific README in that folder to build and deploy.

Performance Testing Idea

Once both services are deployed to Cloud Run, you can use a load testing tool like hey or wrk to generate traffic.

Example Benchmark:

# Test Node.js App
hey -n 1000 -c 50 https://node-app-url.run.app/

# Test Go App
hey -n 1000 -c 50 https://go-app-url.run.app/

Then, visit the Google Cloud Console > Trace to visualize the latency distributions and see the "waterfall" of spans for specific requests.