WorkflowTCO
Infrastructure Capacity Planner

Self-Hosted n8n Queue Worker, Redis & Postgres Cloud Sizing Modeler

Estimate exact container memory, concurrent queue workers, monthly webhook egress, and engineering maintenance cost across Railway, DigitalOcean, Hostinger KVM, Cloudways, and RunPod GPU sidecars.

Dial In Queue Workers, Payload Size & Monthly Ops Hours

Compare raw VPS economics against managed PaaS and serverless GPU enrichment sidecars.

Presets:
workers
13 workers16
GB/mo
50400 GB/mo5,000
hrs/mo
02 hrs/mo12
Optimal host
Cloudways
$154/mo
3-year surplus
$31,306
Minus porting
Horizon
32 days
Recovery span

Multi-Year Ownership Trajectory

Enterprise Cloud Workflow Seat Bundle12m: $12,600 · 36m: $37,800
Cloudways12m: $1,848 · 36m: $5,544 · 32d payback
Railway12m: $1,884 · 36m: $5,652 · 29d payback
Hostinger VPS12m: $2,058 · 36m: $6,174 · 43d payback
RunPod12m: $2,268 · 36m: $6,804 · 51d payback
DigitalOcean12m: $2,292 · 36m: $6,876 · 39d payback

Verified Expenditure & Breakeven Schedule

ProviderBase + UsageHidden Overages12-Month TCO36-Month TCOPaybackRates
CloudwaysLowest 36m
24/7 managed OS security patching cuts agency SRE load
$154/mo$0/mo
None triggered
$1,848$5,54432d ($950)Details →
Railway
Zero Linux patching + 90-second Redis queue blueprint
$157/mo$0/mo
None triggered
$1,884$5,65229d ($850)Details →
Hostinger VPS
Lowest cash outlay ($10.99/mo KVM4) with 16 TB included bandwidth
$171/mo$0/mo
None triggered
$2,058$6,17443d ($1,250)Details →
RunPod
Zero egress + GPU sidecar for local Whisper/OCR workflow steps
$189/mo$0/mo
None triggered
$2,268$6,80451d ($1,450)Details →
DigitalOcean
Best flat-rate bandwidth ($0.01/GB) for heavy file/JSON webhooks
$191/mo$0/mo
None triggered
$2,292$6,87639d ($1,100)Details →

Verified rate card date: 2026-11-21 · Full RAG equations in /llms-full.txt.

Explicit Unit-Economics Math

Queue-Mode Worker Memory Provisioning Formula

recommended_ram_gb = 1.5_main_webhook + (worker_count * 1.25_gb) + 1.0_redis_postgres

A 3-worker queue cluster needs 1.5 + 3.75 + 1.0 = 6.25 GB RAM -> provision an 8 GB RAM host tier.

Why 2 GB RAM Droplets Crash Under Production PDF & Image Webhooks

Running n8n main, PostgreSQL, and Redis on a tiny 2 GB VPS works for light text webhooks but fails reliably whenever simultaneous binary payloads (invoices, audio files, Base64 images) arrive in parallel. Isolating workers into BullMQ queue mode with 8 GB total RAM stops OOM restarts completely.

Cost Estimation Q&A

When does Railway become more expensive than DigitalOcean for self-hosted n8n?

Because Railway bills $0.10/GB for public outbound transfer compared to DigitalOcean’s 2-5 TB bundled pool and $0.01/GB overage, workloads pushing more than 800 GB/month of public media favor DigitalOcean or RunPod.