Why Smart Router?

Smart Router sits between your application and a fleet of RPC providers. It picks the right one per request, retries the wrong one, fans out when correctness matters, and caches what it can.

Smart Router topology — your app to providers via Smart Router with shared cache

What you get

Cost reduction

A standalone cache absorbs repeat reads — calls against finalised blocks, archive lookups, contract-state reads. Multiple router replicas can share one cache. For high-volume reads, the cache hit rate is often the largest contributor to lower monthly RPC spend.

Reliability

Every relay flows through a configurable failover pipeline. Bad responses retry on a different provider. Slow responses get hedged in parallel against a second provider. Critical reads can require consensus across multiple providers before the response is returned.

Visibility

Prometheus metrics, OpenTelemetry traces, and structured logs cover the full request lifecycle — inbound listener, provider selection, retries, hedging, cache lookups, outbound response.

Built for these use cases

dApp / frontend

Stable RPC for production traffic without locking yourself to one provider. Mix paid and free upstreams. Fall back automatically when one stutters. Use directives from the client to bypass cache or pin a provider when debugging.

Data indexing

Indexers hammer eth_getLogs, debug_traceBlock, and similar heavy reads. Smart Router caches finalised-block reads aggressively and rotates providers based on which one currently serves your block range fastest. Consensus mode catches one-off lying providers before bad data hits your index.

Self-hosted RPC infrastructure

A drop-in routing layer in front of your own RPC nodes plus paid providers. Same engine across dev, staging, and production. Same config shape on bare metal or Kubernetes.

What's distinctive

	A reverse proxy	Smart Router
Routing	round-robin / least-conn	QoS-weighted (latency, sync, availability) per method category
Bad provider	retry on the same connection	rotate to a different provider
Slow provider	wait	hedge in parallel against a second provider
Disagreeing providers	last response wins	require consensus across N providers
Caching	URL-keyed, TTL	block-aware, reorg-safe, JSON-RPC native
Heavy methods (`eth_getLogs`, `debug_*`)	treated identically	routed only to capable upstreams (archive, bundler, …)

Beyond EVM

Most routing layers in the ecosystem are EVM-only. Smart Router serves REST, gRPC, and Tendermint RPC alongside JSON-RPC, with first-class support for the Cosmos ecosystem (Lava, Cosmos SDK, IBC, Tendermint). One router, one config shape, every chain you care about. See Supported chains.