The alternatives, honestly

The Graph is one of several ways to get indexed chain data, and it is the right answer less often than a site like this usually admits. Here is what the others are for.

Checked against Graph Horizon (2025-12-11)

Last read 2026-08-30 Due again 2026-11-30

Every protocol claim below was read at these sources on 2026-08-30. Where they disagree with each other, the entry says so.

Every other entry here assumes you have decided to use The Graph. This one does not, because a library that only argues one side is a brochure.

Five shapes, not five products

A decentralised network with verifiable answers. The Graph. Many independent operators, proofs of indexing, disputes and staked capital behind correctness.

Managed subgraph hosting. You write the same subgraph artefact and one company runs it, with a contract and support. Goldsky and Alchemy’s subgraph products sit here.

A self-hosted framework. You run the indexer. Ponder describes itself as “an open-source TypeScript framework for EVM data indexing”; nuthatch is the single-binary version of the same idea.

A hosted data layer. Somebody else has already indexed the chains and sells you access. SQD positions as “the onchain data layer”, covering “blockchain data across 130+ networks”. Envio calls itself “the data layer for blockchain apps”.

An analytics warehouse. Dune and its kin: SQL over pre-decoded chain data, aimed at exploration and dashboards rather than at serving an application.

The comparison

Rows where The Graph loses come first, because a table you win outright is not a comparison.

The GraphManaged hostingSelf-hostedHosted data layerAnalytics warehouse
Working within an hournoyespartlyyesyes
One company to call when it breaksnoyesnoyesyes
Ad-hoc SQL explorationnonoyespartlyyes
No indexing step before you are servednoyesyesyesyes
No third party in the request pathnonoyesnono
Answer verifiable by someone who does not trust the operatoryesnononono
Survives its vendor losing interestyesnoyesnono
Someone else pays for the machinesyesyesnoyesyes

The Graph loses five of eight. That is the honest position, and the two rows it wins are not small: nothing else in the table can produce an answer a stranger can check, and nothing else except running it yourself outlives the company behind it.

Choosing, plainly

Use a managed host or a hosted data layer when you need data this week, your users would not care if the provider changed, and losing the provider would be an inconvenience rather than an existential problem. Most applications are here, and pretending otherwise does nobody a service.

Use an analytics warehouse when the question is exploratory. If you are writing SQL to find out something, rather than serving the same query ten thousand times a day, an application-serving index is the wrong tool and The Graph’s whole machinery is overhead.

Self-host when nothing may sit in the request path, when the subgraph you need does not exist and nobody will index it, or when the data is for you and nobody else has to trust it.

Use The Graph when at least one of these is true, and be honest that if none of them is, you are paying for properties you do not need:

  • Somebody who does not trust you has to be able to check the answer. This is the one nothing else on the list can do at all.
  • Your application must outlive any particular company, including the one you would otherwise buy from.
  • You want redundancy across operators rather than one provider’s uptime.
  • You are contributing to public infrastructure and consider that worth something. Not an engineering argument, and a legitimate reason.

Where the comparison is unfair to The Graph

Two rows deserve a caveat rather than a footnote.

“Working within an hour” is measured against writing and publishing a subgraph. If your question is a token question, the Token API answers it about as fast as anything else on the table, and that is The Graph too.

“No indexing step before you are served” assumes a new subgraph. Querying one that already exists and is well served is immediate, and there are a great many of those.

Where it is unfair to the others

The managed hosts run the same subgraph artefact. Building on a subgraph is not a lock-in decision in the way it usually would be: the same definition can move between a managed host and the network, which materially lowers the cost of starting on one and moving to the other.

Anyone telling you that choice is irreversible has an interest in you thinking so.