homelab-optimized/docs/infrastructure/resource-allocation.md

Resource Allocation Guide

CPU, memory, and storage recommendations for homelab services

---

Overview

This guide provides resource allocation recommendations for services running in the homelab. Values are based on typical usage and should be adjusted based on actual usage patterns.

---

Host Capacity

Current Resources

Host	CPU	RAM	Storage	Workload
Atlantis	8 cores	32GB	40TB	Media, Vault
Calypso	4 cores	32GB	12TB	Infrastructure
Concord NUC	2 cores	16GB	256GB	Light services
Homelab VM	4 cores	28GB	100GB	Monitoring
RPi5	4 cores	16GB	512GB	Edge

Available Headroom

Host	CPU Available	RAM Available	Notes
Atlantis	2 cores	8GB	~25% headroom
Calypso	1 core	12GB	~37% headroom
Concord NUC	0.5 core	4GB	Limited
Homelab VM	1 core	8GB	~28% headroom
RPi5	2 cores	8GB	~50% headroom

---

Service Resource Guidelines

Infrastructure Services

Service	CPU	Memory	Storage	Notes
Nginx Proxy Manager	0.5	256MB	1GB	Minimal
Authentik	1	1GB	10GB	With PostgreSQL
Prometheus	1	2GB	20GB	Adjust for retention
Grafana	0.5	512MB	1GB	Dashboards
Alertmanager	0.25	128MB	-	Minimal

Database Services

Service	CPU	Memory	Storage	Notes
PostgreSQL	1	1GB	10GB+	Per database
Redis	0.5	512MB	-	In-memory
MariaDB/MySQL	1	512MB	5GB	Legacy services

Media Services

Service	CPU	Memory	Storage	Notes
Plex	2+	2GB	-	Transcoding
Jellyfin	2+	2GB	-	Hardware assist
Sonarr	0.5	256MB	-	Low usage
Radarr	0.5	256MB	-	Low usage
Lidarr	0.5	256MB	-	Low usage
Prowlarr	0.25	128MB	-	Minimal
Bazarr	0.5	512MB	-	Subtitle processing
qBittorrent	1	512MB	-	Upload/download
SABnzbd	0.5	256MB	-	Download

Photo Services

Service	CPU	Memory	Storage	Notes
Immich	2	2GB	100GB+	ML processing
PhotoPrism	2	2GB	100GB+	Optional

Communication Services

Service	CPU	Memory	Storage	Notes
Matrix/Synapse	2	1GB	10GB	Federation
Element	0.5	256MB	-	Web client
Mastodon	2	2GB	20GB	Social
Mattermost	1	1GB	5GB	Team chat
Jitsi	2	2GB	-	Video

Home Automation

Service	CPU	Memory	Storage	Notes
Home Assistant	1	2GB	5GB	Core
Zigbee2MQTT	0.5	256MB	-	MQTT broker
Z-Wave JS	0.5	512MB	-	Z-Wave

Development

Service	CPU	Memory	Storage	Notes
Gitea	1	512MB	5GB	Git hosting
Gitea Runner	1	512MB	-	CI/CD
Portainer	0.5	256MB	-	Management
OpenHands	2	4GB	10GB	AI dev (on-demand)

Productivity

Service	CPU	Memory	Storage	Notes
Paperless-NGX	1	1GB	50GB	Document OCR
Wallabag	0.5	256MB	5GB	Read later
Reactive Resume	0.5	256MB	1GB	Resume builder
Seafile	2	2GB	100GB+	File sync

Security

Service	CPU	Memory	Storage	Notes
Vaultwarden	1	512MB	1GB	Passwords
Bitwarden	2	1GB	5GB	(if using official)

Privacy

Service	CPU	Memory	Storage	Notes
Invidious	1	1GB	-	YouTube frontend
Piped	1	1GB	-	Music frontend
Libreddit	0.5	256MB	-	Reddit frontend

DNS & Network

Service	CPU	Memory	Storage	Notes
Pi-hole	0.5	256MB	2GB	DNS filtering
AdGuard	1	512MB	2GB	DNS + ads
WireGuard	0.25	128MB	-	VPN
Headscale	0.5	256MB	-	WireGuard server

---

Memory Limits by Host

Atlantis (32GB)

System:                    4GB
Container overhead:       4GB
Vaultwarden:            512MB
Immich:                 2GB
Plex:                   2GB
ARR stack:              1GB
Jitsi:                  2GB
Matrix:                 1GB
Mastodon:               2GB
Misc services:          2GB
---------------------------
Reserved:              ~15GB

Calypso (32GB)

System:                    4GB
Docker overhead:           4GB
Authentik:               1GB
NPM:                     256MB
Prometheus:              2GB
Grafana:                 512MB
PostgreSQL:              1GB
ARR stack:              512MB
Other services:          3GB
---------------------------
Reserved:              ~16GB

Concord NUC (16GB)

System:                    2GB
Docker:                    2GB
Home Assistant:            2GB
AdGuard:                512MB
Plex:                    2GB
Other services:           2GB
---------------------------
Reserved:               ~5.5GB

---

CPU Limits by Service

High CPU (2+ cores)

• Plex/Jellyfin (transcoding)
• Immich (ML processing)
• OpenHands
• Ollama
• Video processing

Medium CPU (1 core)

• Databases (PostgreSQL, MariaDB)
• Matrix/Synapse
• Mastodon
• Seafile
• Paperless-NGX (OCR)

Low CPU (<1 core)

• Nginx Proxy Manager
• Authentik
• Pi-hole/AdGuard
• Vaultwarden
• Arr suite (Sonarr, Radarr)
• Prometheus (scraping)

---

Storage Guidelines

Media Storage

• Movies/TV: On Atlantis, shared via NFS/SMB
• Music: Dedicated volume
• Photos: Immich primary on Atlantis, backup on RPi5

Application Data

• Prometheus: SSD required (fast writes)
• Databases: SSD required
• Cache: Can be small/fast

Backup Storage

• Local: Dedicated volume on Calypso
• Remote: Backblaze B2 / cold storage

---

Docker Compose Examples

Memory Limits

services:
  prometheus:
    image: prom/prometheus
    deploy:
      resources:
        limits:
          memory: 2G
        reservations:
          memory: 1G

CPU Limits

services:
  plex:
    image: plexinc/pms-docker
    deploy:
      resources:
        limits:
          cpus: '2.0'

---

Monitoring Resource Usage

Check Current Usage

# All containers
docker stats --no-stream

# Specific host
curl http://<host>:9100/metrics | grep node_memory_MemAvailable

# Grafana dashboard
# Infrastructure → Host Resources

Alerts

Metric	Warning	Critical
CPU	>70%	>90%
Memory	>80%	>95%
Disk	>80%	>90%

---

Optimization Tips

1. Use :latest sparingly - Pin versions for stability
2. Enable GPU transcoding - For Plex/Jellyfin
3. Use SSD for databases - Prometheus, PostgreSQL
4. Limit concurrent transcode - In Plex settings
5. Enable Prometheus targerhs - For better monitoring

---

Capacity Planning

Growth Projections

Service	Current	6 Months	12 Months
Media storage	20TB	25TB	30TB
Photo storage	500GB	750GB	1TB
Prometheus	10GB	15GB	20GB
Database	5GB	7GB	10GB

Warning Signs

• Disk usage >80% sustained
• Memory pressure alerts daily
• Container restarts increasing
• CPU throttling visible

---

Links

• Grafana Dashboards
• Docker Guide
• Monitoring Architecture