Beta

Adaptive RPC Load Balancer

by Zeusfyi
4Supported chains 4Plans
Starting at $29.00

Additional credit usage.
Add-on may consume additional plan credits when used.

This add-on is currently in Beta

Feel free to test it, we would appreciate feedback to address any errors that may occur

Load balance RPC volume at scale across QuickNode, other infrastructure RPC providers, and even your own endpoints if you run your own infrastructure.

Top Reason to Use the Adaptive RPC Load Balancer?

There's many, but the top one? People are sick of hitting 429 rate limiting and 5xx errors. What's double worse is paying for the request that failed. So we set out to solve this problem.

- 5 endpoints limited to 100/reqs -> 1 endpoint that handles 500 req/s

Now there's a solution, backed by extensive studies. Turn many endpoints into one super endpoint that handles the scale you need without the errors. It can handle Nx more requests with N being the number of requests/sec than any single endpoint can handle. That's in addition to the other significant proven performance gains and error rate reductions you can expect from our adaptive load balancing technology, see our benchmarking section for details on our docs page.

Serverless EVM Execution Environments
- Using Foundry's Anvil

Overview:
a. Each serverless environment lasts for up to 10 minutes, but can be released early
b. Each environment contains an isolated ephemeral anvil service
c. Run up to 5 concurrent serverless execution environments for free.
d. Continue to run simulations at scale post beta at 250 compute units per session. Which is 10x cheaper than Tenderly or Blocknative.

Free Access During Beta Testing
- Serverless EVMs
- Enterprise grade real-time mempool service

Getting Started

Prerequisites Setup


You'll need to get an API key from us to use the service. If you login to the dashboard via the SSO link you should navigate to Access on the bottom lefthand navigation panel and then request a new API key. You can also configure every aspect of the load balancer via Dashboard UI, or purely via API.

Adaptive Load Balancing Procedure


You then use the name of your route table group as a header like the below example.

Procedure Function Routing


To use embedded protocol procedures you only need to add the key value to your payload. In this example, to use the max block procedure for Ethereum, which polls your routing table for the current block number, and then forwards your request to the endpoints returning the highest block number seen and then returns the first successful response.

Curl Example: Procedures


"procedure": "eth_maxBlockAggReduce"


curl --location 'https://iris.zeus.fyi/v1/router' \
--header 'Content-Type: application/json' \
--header 'Authorization: Bearer YOUR-BEARER-TOKEN' \
--header 'X-Route-Group: ethereum-mainnet' \
--data '{
    "jsonrpc": "2.0",
    "procedure": "eth_maxBlockAggReduce",
    "method": "eth_getBlockByNumber",
    "params": ["latest", true],
    "id": 1
}'

Supported Chains

Arbitrum
Avalanche
Base
Ethereum logo
Ethereum

Published on Sep 8, 2023

Support Website Support Email

Plans & features

Discover - $29/month
  • Discover - $29/month
  • Lite - $99/month
  • Standard - $299/month
  • Performance - $999/month
Enterprise Sales
5M requests. 500 requests/second, 99% uptime SLA
Tunable Adaptive RPC Load Balancer
Our load balancer comes with adaptive load balancing that’s proven to help shape load balancing distribution in a highly efficient way over a large possible range of scenarios in real-time using t-digest driven latency statistics. Studies show median latency improvement by 15-18%, and reduction in error rates by 50% when compared to the round robin alternative in many real world cases, while also making your super endpoint as N x req/s with N being each table endpoint.
Turns many endpoints into one super endpoint
All QuickNode users will have their endpoints automatically registered with the load balancer, and will have automatically generated routing group tables based on the network and chain type for that endpoint for all plans. E.g. ethereum mainnet endpoints will be automatically registered to the ethereum-mainnet routing group table. These won't count against your routing table limits unless you make any manual changes to them.
Serverless EVMs for Tx Simulations & Forking
Each serverless environment lasts for up to 10 minutes, but can be released early. Each environment contains an isolated ephemeral anvil service. Uses your load balanced own endpoints for forking live networks. Post beta, will only be 250 ZU compute units per serverless session, which is ~10x cheaper than anywhere elsewhere. They're also lightning fast, using our own serverless in-house technology.
Latency Performance Dashboard, UI & APIs
Use a single endpoint and only change your header to switch between groups of nodes categorized for specific purposes like indexing, or between network and chain options.
Pipeline Requests with Procedures
We've included embedded routing procedures for the protocols we support that will poll your table routes for the latest block number, and then send the request to routes at the max block number seen and then return the first successful response. We're also evaluating new ones, and planning to open the UI to allow you to build them yourself.
Extensive Documentation & Support
Get help with questions on Discord, email, or schedule a demo. We also have an extensive documentation site at https://docs.zeus.fyi/docs/lb/intro