Lab 05: The POC Sprint

Learning Objectives

By completing this lab, you will:

• Scope a time-boxed proof of concept effectively
• Define measurable success criteria
• Deploy minimal viable infrastructure quickly (Kind, GCP, or AWS)
• Prepare and deliver effective demos
• Document outcomes for stakeholders
• Handle common POC challenges and questions

Prerequisites

Common Prerequisites

• Terraform >= 1.5 (for cloud deployments)
• kubectl installed
• Helm 3.x installed
• Basic understanding of Kubernetes concepts

GCP Prerequisites (if using GCP)

• GCP project with billing enabled
• gcloud CLI configured

AWS Prerequisites (if using AWS)

• AWS account with appropriate permissions
• aws CLI configured

Local Prerequisites (if using Kind)

• Docker installed
• Kind installed (brew install kind)

What is a POC?

A Proof of Concept (POC) is a time-boxed demonstration that validates whether a solution works in a customer's environment. Unlike a full implementation, a POC:

• Focuses on core functionality - Not every feature
• Uses minimal infrastructure - Fast to deploy, easy to clean up
• Has clear success criteria - Everyone knows what "success" means
• Is time-boxed - Usually 1-4 weeks
• Produces a decision - Go/no-go for full implementation

Deployment Options

This lab supports three deployment options for POCs:

Option 1: Kind (Local, Zero Cost) ⭐ Recommended for Learning

Best for: Learning, testing, zero-cost POCs

Pros:

• ✅ Zero cost
• ✅ Fastest deployment (~2 minutes)
• ✅ No cloud account needed
• ✅ Perfect for local testing

Cons:

• ❌ Not representative of real cloud environment
• ❌ Limited to local machine resources

Quick Start:

# Just run the script - it will detect no terraform.tfvars and use Kind
./scripts/quick-deploy.sh

Option 2: GCP (Cloud, Minimal Cost)

Best for: Real POCs in GCP environments

Pros:

• ✅ Real cloud environment
• ✅ Fast deployment (~5-10 minutes)
• ✅ Minimal cost (~$0-5/day)
• ✅ Easy cleanup

Cons:

• ❌ Requires GCP account
• ❌ Small cost

Quick Start:

cd minimal-deployment
cp terraform.tfvars.example terraform.tfvars
# Edit terraform.tfvars: set cloud_provider = "gcp" and project_id
cd ..
./scripts/quick-deploy.sh

Option 3: AWS (Cloud, Minimal Cost)

Best for: Real POCs in AWS environments

Pros:

• ✅ Real cloud environment
• ✅ Fast deployment (~10-15 minutes)
• ✅ Minimal cost (~$0-5/day)
• ✅ Easy cleanup

Cons:

• ❌ Requires AWS account
• ❌ Small cost

Quick Start:

cd minimal-deployment
cp terraform.tfvars.example terraform.tfvars
# Edit terraform.tfvars: set cloud_provider = "aws" and region
cd ..
./scripts/quick-deploy.sh

Quick Start

Recommended: Use Quick Deploy Script

The quick-deploy.sh script automatically detects your deployment method:

cd labs/05-poc-sprint

# If no terraform.tfvars exists, it will use Kind (zero cost)
# If terraform.tfvars exists, it will use the cloud provider specified
./scripts/quick-deploy.sh

Manual: Cloud Deployment

cd minimal-deployment
cp terraform.tfvars.example terraform.tfvars
# Edit terraform.tfvars with your cloud provider settings

# Deploy
terraform init
terraform apply

# Get credentials
terraform output get_credentials_command
# Run the output command

# Deploy Argo Workflows
helm repo add argo https://argoproj.github.io/argo-helm
helm install argo-workflows argo/argo-workflows --namespace argo --create-namespace

Prepare Demo

./scripts/prepare-demo.sh

Lab Structure

labs/05-poc-sprint/
├── templates/              # POC planning templates
│   ├── poc-scope-document.md
│   ├── success-criteria.md
│   ├── daily-standup-format.md
│   └── final-report-template.md
├── minimal-deployment/     # Fast infrastructure deployment
│   ├── main.tf            # Supports GCP and AWS
│   ├── variables.tf
│   └── terraform.tfvars.example
├── manifests/              # Demo workflows
│   ├── demo-workflow-simple.yaml
│   ├── demo-workflow-multistep.yaml
│   └── demo-workflow-parallel.yaml
├── demo-prep/              # Demo preparation materials
│   ├── demo-script.md
│   ├── backup-demo.md
│   └── common-questions.md
└── scripts/                # Automation scripts
    ├── quick-deploy.sh     # Supports Kind, GCP, and AWS
    ├── prepare-demo.sh
    └── cleanup.sh

Key Deliverables

1. POC Scope Document

Use templates/poc-scope-document.md to define:

• Objectives (primary and secondary)
• Scope (in and out of scope)
• Success criteria
• Timeline
• Resources needed

2. Success Criteria Framework

Use templates/success-criteria.md to define:

• Must-have criteria (POC fails without these)
• Should-have criteria (important but not blocking)
• Nice-to-have criteria (if time permits)

3. Demo Materials

• Demo Script: Step-by-step demo flow (demo-prep/demo-script.md)
• Backup Demo: Plan B if live demo fails (demo-prep/backup-demo.md)
• Common Questions: Prepared answers (demo-prep/common-questions.md)

4. Final Report

Use templates/final-report-template.md to document:

• What was accomplished
• Success criteria results
• Lessons learned
• Recommendations
• Next steps

POC Best Practices

Scoping

✅ Start Small: Focus on core functionality ✅ Be Specific: Clear objectives and success criteria ✅ Time-Box: Set a deadline and stick to it ✅ Document Everything: Scope, decisions, outcomes

Execution

✅ Deploy Quickly: Use minimal infrastructure ✅ Test Early: Validate assumptions quickly ✅ Communicate Often: Daily standups, regular updates ✅ Stay Focused: Resist scope creep

Demo

✅ Prepare Thoroughly: Test everything beforehand ✅ Have Backup Plan: Screenshots, videos, architecture discussion ✅ Engage Audience: Ask questions, check understanding ✅ Be Honest: Acknowledge limitations

Documentation

✅ Document Outcomes: What worked, what didn't ✅ Capture Lessons: What would you do differently? ✅ Provide Recommendations: Clear next steps ✅ Share Knowledge: Help others learn

Deployment Comparison

Feature	Kind	GCP	AWS
Cost	$0	~$0-5/day	~$0-5/day
Deployment Time	~2 min	~5-10 min	~10-15 min
Setup Complexity	Low	Medium	Medium
Real Cloud Environment	❌	✅	✅
Best For	Learning	GCP POCs	AWS POCs

When to Use Each

Kind (Local):

• Learning and testing
• Zero-cost requirement
• Fast iteration
• No cloud account needed

GCP:

• Customer uses GCP
• Need real cloud environment
• Quick POC deployment
• Minimal cost acceptable

AWS:

• Customer uses AWS
• Need real cloud environment
• Quick POC deployment
• Minimal cost acceptable

Estimated Time

Infrastructure Deployment:

• Kind: ~2 minutes
• GCP: ~5-10 minutes
• AWS: ~10-15 minutes

Demo Preparation: 30-60 minutes POC Planning: 1-2 hours (using templates) Total Lab Time: 2-3 hours

Estimated Cost

Kind Option: $0 (local deployment) GCP Option: $0-5 per day (minimal resources, destroy quickly) AWS Option: $0-5 per day (minimal resources, destroy quickly)

Cost Breakdown (Cloud):

• Cluster: ~$0.10/hour
• Nodes: ~$0.05-0.10/hour per node
• Load balancer: ~$0.025/hour
• Total: ~$0.20-0.30/hour = ~$5-7/day if running 24/7

Tip: Destroy resources immediately after POC to minimize costs.

Documentation

• Scoping Guide - How to scope a POC effectively
• Demo Guide - Comprehensive demo preparation and delivery
• Troubleshooting - Common POC issues and solutions
• Step-by-Step Guide - Detailed walkthrough

Templates

• POC Scope Document
• Success Criteria Framework
• Daily Standup Format
• Final Report Template

Demo Materials

• Demo Script
• Backup Demo Plan
• Common Questions

Cleanup

Kind

kind delete cluster --name poc-cluster

Cloud (GCP or AWS)

./scripts/cleanup.sh

Or manually:

cd minimal-deployment
terraform destroy

Important: Always clean up POC resources immediately to minimize costs!

Next Steps

After completing this lab:

1. Practice scoping a POC using the templates
2. Prepare a demo using the demo materials
3. Review the common questions and prepare answers
4. Try deploying with different providers (Kind, GCP, AWS)
5. Proceed to Lab 06: Multi-Tenant Deployment

Additional Resources

• POC Best Practices
• Demo Best Practices
• Stakeholder Communication