If you like eRPC, give it a star on GitHub ⭐️
Why eRPC?

Why eRPC?

These are the main reasons eRPC was built:

  • To reduce overall costs of RPC usage and egress traffic, by local caching.
  • To provide a fault-tolerant and reliable source for RPC consumers in case of one or more provider outages.
  • To give a global observability about your RPC usage both within internal teams and projects, and towards the upstream RPC 3rd party companies.

Features

  • Failover across multiple upstreams by tracking their response time, error rate, blockchain sync status, etc.
  • Offer self-imposed rate limitting per project, or network, or upstream to avoid abuse, and unintentional DDoS.
  • Prometheus metrics collection and Grafana dashboards to monitor costs, usage and health of your RPC endpoints.
  • Caching and multiplexing (auto-merging identical requests) to reduce redundant RPC calls and costs.

Use-cases

For the start the main focus of eRPC will be on read-heavy use-cases such as:

Frontend of dApps

Often many requests made from dApps can be multiplexed into a single RPC call, or use cache to reduce costs.

Here are real production case studies where eRPC serves 8k RPS and 1B requests / month:

Data indexing using Ponder, Graph, Envio, Flair, etc.

When using any of the web3 indexing tools you will be making lots of requests towards your RPC provider, especially during re-backfills for historical data.

eRPC can help in two main areas:

  • Cache already made RPC calls (eth_getLogs, eth_call, eth_getBlockByNumber, etc)
  • Rate-limit upstream pressure towards RPC nodes to avoid fatal error

Resilient load balancer for self-hosted RPC nodes

For app-specific chains and rollups, or projects who prefer to self-host RPC nodes for performance reasons, it makes sense to use eRPC as the entry-point load-balancer.

Compared to more traditional LB solutions (ALB, K8S Services, etc) eRPC will provide EVM-focused features like:

  • EVM-aware healthchecks (e.g. how many blocks behind)
  • EVM-aware fallbacks (e.g. if a 4xx is because of missing block, try another upstream)
  • EVM-aware method filters (e.g. certain methods to go to node A and other methods to go to node B)
Quick startFAQ