Do I need to run a node to index blockchain data?

Updated August 2026

No. Hosted endpoints give you indexed blockchain data — full history and live streams — without operating any infrastructure. You get an API key and start consuming data.

Running your own nodes is a legitimate choice, but it is a decision about control, cost at scale, and compliance. It is not a prerequisite.

What are the options?

	Hosted endpoint	Self-hosted
Setup	API key	Node operation, storage, monitoring
Ongoing effort	None	Real, continuous
Upfront cost	Free tier available	Hardware and engineering time
Cost at very large scale	Grows with usage	Can be lower
Control over data	Vendor-operated	Complete
Suits	Almost everyone starting out	Scale, compliance, or specific requirements

What does self-hosting actually involve?

More than "run a node." To produce indexed data yourself you need an instrumented node emitting execution data, a reader capturing it, a merger consolidating blocks into files, storage for those files, and a server delivering streams to consumers. Then monitoring, alerting, upgrades, and someone on call when a chain hard-forks.

This is a real infrastructure commitment. It is also entirely possible — Firehose and Substreams are open source under Apache 2.0, and the whole stack can be self-hosted with no licensing restrictions and no vendor permission required.

When is self-hosting the right call?

Scale. At sufficient volume, operating infrastructure becomes cheaper than paying per usage. Where that crossover sits depends on your workload.

Compliance or data residency. Some organisations cannot send queries to third-party infrastructure, or need data held in a specific jurisdiction.

Chains nobody hosts. If you need a network no provider supports — including your own — self-hosting is the only path.

Avoiding lock-in. Because the stack is open source, self-hosting is a genuine exit option rather than a theoretical one. That matters more than it sounds: the ability to leave is what keeps a hosted relationship honest, and it does not exist with closed platforms.

What's the sensible default?

Start hosted. Get your pipeline working, learn what your actual data requirements are, and find out whether the volumes justify infrastructure. Self-hosting a stack to serve a workload you have not characterised yet is a common and expensive mistake.

Because the same software runs in both places, moving later is a deployment change rather than a rewrite.

Frequently asked questions

Is the free tier enough to evaluate? Yes. An API key from thegraph.market requires no personal information and is enough to build and test a pipeline.

Can I self-host later without rewriting? Yes. The same open-source components run in both modes, so your modules and sinks are unchanged.

Do I need an archive node? Usually not — see do I need an archive node?

Start without infrastructure — get an API key at thegraph.market, no personal information required. Or self-host from github.com/streamingfast.