Why Node Runners?
If you need to run blockchain nodes on AWS, you have several options. This page explains when Node Runners is the right choice — and when it isn't.
The Problem
Running blockchain nodes looks simple until you try it at scale:
- Each protocol is different. Solana needs 512 GB RAM and NVMe RAID. Ethereum needs separate execution and consensus clients. Bitcoin is CPU-light but storage-heavy. There's no universal "just deploy a node" button.
- Instance selection is non-obvious. The wrong EC2 instance type means either overpaying 2–3× or nodes that can't keep up with chain tip. Figuring out the optimal configuration requires protocol-specific benchmarking that few teams have time for.
- Operational burden compounds. Disk growth, snapshot management, client upgrades, healthchecks, key rotation — each node type has its own failure modes. Multiply by 5+ protocols and it becomes a full-time job.
- AI assistants can't help without structure. LLMs have general knowledge about blockchain nodes, but without a well-defined infrastructure framework, they can't reliably generate correct CDK code or make cost-aware recommendations.
How Node Runners Compares
| Node Runners | Managed RPC | DIY on EC2 | Docker Compose / K8s | |
|---|---|---|---|---|
| Cost control | Full — you own the EC2, pay only compute + storage | Per-request pricing, opaque at scale | Full, but manual optimisation | Full, but cluster overhead |
| Data sovereignty | Runs in your VPC, your account | Third-party infrastructure | Your account | Your account |
| Protocol coverage | 6 built-in (5 protocols + Dummy reference) + community blueprints | Broad (20+ chains) | Any (manual effort per chain) | Any (manual effort per chain) |
| Time to deploy | Minutes to provision infra, AI-guided or CLI (chain sync time varies by protocol) | Minutes (API key) | Hours to days | Hours |
| Customisation | Full — modify blueprints, instance types, networking | Limited | Unlimited but unstructured | Unlimited but complex |
| AI-assisted operations | Built-in — deploy, healthcheck, cost estimate via natural language | Not applicable | Not available | Not available |
| Operational automation | CloudWatch agent pre-configured for logs and metrics; alarms and snapshots are documented but user-configured | Managed by provider | Build it yourself | Build it yourself |
| Multi-protocol consistency | Same CDK framework, same commands, same monitoring for all chains | N/A — different API per provider | Different scripts per chain | Different charts per chain |
| Cost visibility | Static per-protocol cost tables, with optional real-time AWS Pricing API breakdown | Invoice after the fact | Manual calculation | Manual calculation |
When to Use Node Runners
✅ Use Node Runners when you:
- Need self-hosted nodes in your own AWS account (compliance, data sovereignty, latency control)
- Want to experiment with multiple protocols without rebuilding infrastructure from scratch
- Prefer AI-guided workflows over reading protocol-specific documentation
- Care about cost optimisation and want pre-deployment cost estimates
- Need production-like HA setups for evaluation or staging environments
- Want a repeatable, version-controlled infrastructure (CDK) rather than click-ops
When NOT to Use Node Runners
❌ Consider alternatives when you:
- Just need RPC access and don't care where nodes run → use a managed provider (QuickNode, Alchemy, Infura)
- Need 50+ dedicated nodes across 20 protocols with SLA → you likely need a dedicated infrastructure team and custom automation
- Are running validators for profit and need sub-second failover → Node Runners is designed for experimentation and evaluation, not for operating commercial validator businesses
- Need a protocol not yet supported and don't want to create a blueprint → check the community blueprints catalog or request one
The AI Advantage
What makes Node Runners different from "just CDK code" is the AI-first design:
- Describe, don't configure. Tell the AI assistant "I want to run a Solana RPC node optimised for cost" — it recommends the instance type, storage config, and networking setup.
- Cost-aware on request. Before deployment, the AI shows the per-protocol cost estimate from the blueprint, and on request runs a detailed real-time breakdown via the AWS Pricing API alongside performance trade-offs.
- Structured for LLMs. The blueprint system gives AI assistants a well-defined schema to work with — no hallucinated CDK constructs, no guessing at configuration shapes.
- Operational continuity. After deployment, the same AI workflows help with healthchecks, performance diagnosis, and protocol upgrades.
Get Started
Ready to try it? Head to the Quickstart to deploy your first node in under 10 minutes.