EC2 Tier Recommendations

Server sizing guide for the GSign platform.

Memory Requirements

Full Infrastructure

Component	Memory
GitLab CE	4 GB
PostgreSQL	512 MB
MongoDB	512 MB
Redis	256 MB
Prometheus	512 MB
Grafana	256 MB
Loki	512 MB
Go Backend	256 MB
Next.js Frontend	256 MB
HSM Backend	128 MB
Documentation	64 MB
System (Ubuntu + NGINX)	512 MB
Total	~9.9 GB

Recommended Tiers

t3.large (Recommended)

Spec	Value
vCPU	2
RAM	8 GB
Network	Moderate
Cost	~$60/month

Best for:

• Full infrastructure deployment
• Comfortable headroom
• Development + staging

t3.medium (Budget)

Spec	Value
vCPU	2
RAM	4 GB
Network	Moderate
Cost	~$30/month

Requires:

• Disable GitLab Runner (saves 1 GB)
• Use Loki instead of Elasticsearch
• Remove database replicas
• Reduce Prometheus retention

t3.xlarge (Performance)

Spec	Value
vCPU	4
RAM	16 GB
Network	High
Cost	~$120/month

Best for:

• Team development
• High traffic staging
• Additional services

AWS Lightsail Alternative

Lightsail 8 GB (Best Value)

Spec	Value
vCPU	2
RAM	8 GB
SSD	160 GB
Transfer	4 TB
Cost	$40/month

Benefits:

• Fixed pricing (no surprises)
• Same performance as t3.large
• Simpler management
• 4 TB transfer included

Cost Optimization

Spot Instances

Save up to 70% with spot instances:

Instance	On-Demand	Spot
t3.large	$60/month	~$18/month
t3.xlarge	$120/month	~$36/month

Risk: Instance can be interrupted (2-minute warning)

Use case: Development environments, batch processing

Reserved Instances

Save 40-60% with 1-3 year commitment:

Term	Discount
1 year, no upfront	40%
1 year, all upfront	45%
3 year, all upfront	60%

Stop When Not in Use

Development server running 8 hours/day:

• Regular: $60/month
• 8 hours/day: ~$20/month

Weekend shutdown:

• Regular: $60/month
• Weekdays only: ~$45/month

Lightweight Setup (4 GB)

For t3.medium with 4 GB RAM:

Disable GitLab Runner

# Remove from docker-compose
# gitlab-runner: ...

Saves: 1 GB

Use Loki Instead of EFK

Already using Loki (lighter than Elasticsearch)

Saves: 3 GB (vs EFK stack)

Single Database Instances

No replicas:

• PostgreSQL: 1 instance
• MongoDB: 1 instance
• Redis: 1 instance

Saves: 800 MB

Reduce Prometheus Retention

# prometheus.yml
--storage.tsdb.retention.time=7d  # from 30d

Saves: 200 MB

Result: 9.9 GB → ~5.2 GB (fits in 4 GB with swap)

Quick Decision Matrix

Requirement	Recommendation
Full infrastructure, comfort	t3.large ($60)
Budget development	t3.medium ($30) + optimizations
Team development	t3.xlarge ($120)
Best value	Lightsail 8 GB ($40)
Production preview	t3.large or Lightsail
CI/CD with builds	t3.large minimum

Setup Commands

EC2 t3.large

# Launch instance
aws ec2 run-instances \
  --instance-type t3.large \
  --image-id ami-0xxx \
  --key-name gsign-key \
  --security-group-ids sg-xxx \
  --block-device-mappings "[{\"DeviceName\":\"/dev/sda1\",\"Ebs\":{\"VolumeSize\":100}}]"

Lightsail

# Create instance
aws lightsail create-instances \
  --instance-names gsign-dev \
  --availability-zone eu-north-1a \
  --blueprint-id ubuntu_24_04 \
  --bundle-id large_3_0

Monitoring Resources

# Memory usage
free -h

# Disk usage
df -h

# Container resources
docker stats

# Top processes
htop

Disk Recommendations

Environment	Size	Type
Development	50 GB	gp3
Staging	100 GB	gp3
Production	200+ GB	gp3

Note: GitLab + databases can grow quickly. Monitor disk usage.