Skip to main content

Indexer Network (Future)

The current architecture runs a single indexer instance. This is sufficient for launch but introduces a single point of failure for discoverability.

The long-term architecture supports multiple independent indexers.

How It Works​

  • Each indexer watches the same chain events
  • Each produces deterministically identical state
  • Keepers and clients can query any indexer
  • Quorum across indexers provides high availability

Why Multiple Indexers Converge​

Because indexer state is deterministic, any two honest indexers will agree. Disagreement indicates either:

  • one indexer is behind (temporary)
  • one indexer is corrupted (rebuild from chain)

Clients can query multiple indexers and take the majority response. If one indexer is down, others continue serving.

What Quorum Provides​

  • Near 100% uptime for discoverability
  • Redundancy against infrastructure failures
  • Geographic distribution for latency
  • No single operator dependency

What Quorum Does Not Change​

  • Correctness still comes from the chain
  • Indexers still cannot cause invalid payments
  • Indexers still cannot bypass authorization
  • The trust model remains unchanged

Multiple indexers improve availability of the convenience layer. They do not change the security model.

Why Not Required for MVP​

A single indexer is:

  • simpler to operate
  • sufficient for initial volume
  • easier to debug

The cost is a single point of failure for discoverability. At launch scale, this is acceptable. As the protocol grows, indexer redundancy becomes operationally necessary.

The architecture supports this evolution without protocol changes.