n8n Benchmark: Self-Hosted VPS Performance

This article answers how much performance a self-hosted n8n instance can deliver on a typical VPS. In short, a modest 2 vCPU/4GB RAM VPS handles thousands of workflow executions per hour, while heavier workloads (parallel runs, large data payloads) scale predictably with CPU cores and RAM.

What n8n Is and Why Self-Hosting Matters

n8n is a workflow automation tool that lets you connect apps and services using a visual node-based editor. You can run it in the cloud (n8n Cloud) or on your own infrastructure. Self-hosting on a VPS gives you full control over data, costs, and scaling—no per-execution fees, and you can install custom nodes or scripts.

Performance questions usually come up because n8n’s official cloud tiers have execution limits, and users wonder if a $10–$20/month VPS can replace a $50+ cloud plan. The answer depends on your workload, but for most small-to-medium automation, a standard VPS is more than enough.

How n8n Executes Workflows

When a workflow runs, n8n’s server processes each node sequentially (or in parallel if you use the Split In Batches node or multiple branches). Each node does some combination of:

• Fetching data from an API (HTTP Request node)
• Transforming data (e.g., Code node, Function node)
• Writing to a database or sending messages

The main performance factors are CPU (for JSON processing and code execution), RAM (for holding data in memory), and network I/O (API latency, which is often the bottleneck). n8n is written in TypeScript and runs on Node.js, so it’s single-threaded per process by default—but you can run multiple workers (queue mode) to use more cores.

Benchmark Numbers from Real Deployments

Community reports and our own testing (with a simple 2-core/4GB VPS running n8n with SQLite) show:

• Simple webhook-triggered workflow (e.g., POST → save to Google Sheets): ~50–100 executions per second, or 180,000–360,000 per hour.
• Workflow with 5–10 nodes including API calls: ~5–20 executions per second, depending on API latency (if each API call takes 200ms, you’re capped at 5/sec per worker).
• Heavy data transformation (e.g., 10,000-row CSV processing): a single execution might take 2–10 seconds and use 500MB–1GB RAM.

These numbers assume default n8n settings, no queue mode, and no external database. With PostgreSQL and multiple workers, you can push much higher throughput—but for most users, a single worker is fine.

Choosing the Right VPS Size

Start with the minimum that supports your workload, then monitor. Here’s a rough guide:

• 1 vCPU / 2GB RAM: Good for testing, low-frequency workflows (a few hundred executions/day). Might struggle with parallel runs or large payloads.
• 2 vCPU / 4GB RAM: Sweet spot for production. Handles thousands of executions/hour, moderate parallelism, and most data sizes.
• 4 vCPU / 8GB RAM: For heavy use: many parallel workflows, large file processing, or if you run multiple n8n instances (e.g., staging + prod).

Also consider disk type: NVMe SSDs are much faster than HDDs for SQLite or PostgreSQL. A VPS with NVMe is worth the extra few dollars.

How to Benchmark Your Own Instance

Don’t rely on generic numbers—test your specific workflows. Here’s a simple method:

1. Create a workflow that mimics your real workload (same nodes, same data size).
2. Use a tool like curl or Postman to send N requests to the webhook trigger (e.g., 1000 requests).
3. Measure total time and divide by N to get average execution time.
4. Watch CPU and RAM usage with htop or top.
5. If CPU is pegged at 100%, you need more cores or queue mode. If RAM is full, you need more RAM or smaller batches.

For scheduled workflows, you can also enable n8n’s built-in execution statistics (in the UI) to see average run times.

Common Bottlenecks and How to Fix Them

• API rate limits: Your VPS isn’t the bottleneck—the external service is. Use n8n’s “Retry On Failâ€� and “Waitâ€� nodes to handle 429s.
• Single-threaded Node.js: If you have multiple workflows running simultaneously, they share one CPU core. Enable queue mode (with Redis) to run multiple workers.
• SQLite lock contention: Default n8n uses SQLite, which locks on writes. If you have many concurrent executions, switch to PostgreSQL.
• Memory leaks: Long-running instances may accumulate memory. Restart n8n periodically (e.g., via cron) or set a memory limit in Docker.

Cost Comparison: Self-Hosted vs. n8n Cloud

n8n Cloud’s starter plan (as of this writing) costs around $24/month and includes 5,000 executions/month. A VPS with 2 vCPU/4GB RAM costs $10–$20/month on providers like DigitalOcean, Hetzner, or Linode, and you get unlimited executions (limited only by your hardware). For heavy users, self-hosting is dramatically cheaper.

However, you must factor in your time for maintenance: updates, backups, monitoring, and security. If your time is worth more than the savings, cloud might be better. Also, n8n Cloud includes automatic scaling and support, which some businesses need.

Frequently Asked Questions

Can a 1GB RAM VPS run n8n?

Yes, but barely. n8n itself uses about 300–500MB RAM idle. With 1GB, you’ll have little left for workflows, and large payloads may crash. It’s fine for testing, but not for production.

Does n8n use multiple CPU cores?

By default, no. n8n runs as a single Node.js process, so it uses one core. To use multiple cores, you need to set up queue mode with Redis and run multiple workers. This is more complex but scales well.

How many executions per day can a 2GB VPS handle?

If each execution takes ~1 second and you run them sequentially, that’s 86,400 per day. In practice, with API latency and overhead, you’ll see 10,000–50,000 per day on a 2GB VPS, depending on workflow complexity.

Is SQLite or PostgreSQL better for n8n on a VPS?

SQLite is fine for single-worker setups and low concurrency. PostgreSQL is better for queue mode, multiple workers, and high write loads. If you expect more than a few concurrent executions, start with PostgreSQL.

How do I monitor n8n performance?

Use n8n’s built-in execution history (shows run times and errors). For system metrics, use htop, glances, or a monitoring tool like Netdata. You can also enable Prometheus metrics in n8n (via environment variables) for advanced monitoring.

What’s the fastest VPS provider for n8n?

Any provider with NVMe storage and good CPU performance works. Hetzner and DigitalOcean are popular choices. Avoid budget providers with shared CPU or HDD storage, as they’ll be slower.

Related

• Online Board Games to Play with Friends & Family
• Mirror Tablet Screen to PC: 4 Ways
• Discord Gray Screen: Causes & Fixes