RPC Providers

Q: What about geographic latency?

Choose region closest to users and servers; test getSlot RTT from deployment region.

Choose between managed RPC providers (Helius, Triton, QuickNode) and self-hosted Agave validators based on traffic, features, and operational cost.

Recipe

Quick-reference recipe card - copy-paste ready.

import { createSolanaRpc } from "@solana/kit";
 
// Provider (API key in URL or header per vendor)
const rpc = createSolanaRpc(
  "https://mainnet.helius-rpc.com/?api-key=YOUR_KEY"
);
 
await rpc.getSlot().send();

# Self-hosted (operator)
agave-validator --ledger /data/ledger --rpc-port 8899

When to reach for this:

Mainnet or devnet apps exceeding public RPC rate limits.
Needing DAS, webhooks, or priority fee estimate APIs.
Running indexers requiring dedicated stake and bandwidth.
Compliance requirements for private RPC ingress.

Working Example

import { createSolanaRpc, createSolanaRpcSubscriptions } from "@solana/kit";
 
const HTTP = process.env.RPC_HTTP_URL!;
const WSS = process.env.RPC_WSS_URL!;
 
const rpc = createSolanaRpc(HTTP);
const rpcSubscriptions = createSolanaRpcSubscriptions(WSS);
 
const slot = await rpc.getSlot().send();
console.log("slot:", slot);
 
// Health check script for CI
if (slot < 1) throw new Error("RPC unhealthy");

What this demonstrates:

Separate HTTP and WSS URLs from provider dashboard.
Environment variables keep API keys out of source control.
Simple getSlot probe validates endpoint health.

Deep Dive

How It Works

Managed providers run Agave validators plus load balancers, caching, and indexer sidecars.
API keys authenticate and meter requests per plan tier.
Self-hosting runs agave-validator with your own stake, ops, and failover.
Feature matrix (DAS, gRPC, webhooks) varies by vendor - not commodity RPC.

Provider Comparison (high level)

Option	Strengths	Trade-offs
Helius	DAS, webhooks, dev tooling	Vendor coupling for advanced APIs
Triton	Yellowstone gRPC, performance	Ops integration complexity
QuickNode	Multi-chain familiarity	Solana-specific feature depth varies
Public RPC	Free, zero setup	Rate limits, no SLA
Self-host	Control, privacy	Staffing, stake, reliability burden

TypeScript Notes

// Failover: try secondary URL on 429/503
async function rpcWithFallback<T>(fn: (rpc: ReturnType<typeof createSolanaRpc>) => Promise<T>) {
  const primary = createSolanaRpc(process.env.RPC_PRIMARY!);
  try {
    return await fn(primary);
  } catch (e) {
    const secondary = createSolanaRpc(process.env.RPC_SECONDARY!);
    return await fn(secondary);
  }
}

Gotchas

Embedding API keys in frontend - keys get scraped from browsers. Fix: proxy reads through your backend for sensitive keys.
Single provider dependency - outage stops your app. Fix: secondary URL failover and cached critical reads.
Using free tier for production mint - 429 during spike. Fix: paid plan sized to peak RPS.
Mismatched HTTP/WSS hosts - subscriptions fail silently. Fix: copy paired endpoints from dashboard.
Self-host without monitoring - disk full on ledger stops RPC. Fix: ledger alerts and redundant nodes.

Alternatives

Alternative	Use When	Don't Use When
Public devnet RPC	Prototypes	Production mainnet
Colocated validator + provider hybrid	Trading firms	Simple NFT mint site
IPFS-only metadata	Static assets	RPC needs
Light client (future/experimental)	Extreme decentralization	Current production default

FAQs

Do I need a provider for devnet?

Public devnet works early; provider improves reliability for team CI and demos.

How do I pick plan size?

Estimate peak RPS from MAU, GPA weight, and websocket connections - load test before launch.

Can one key serve HTTP and WSS?

Usually same API key on paired URLs - verify vendor docs.

Is self-host cheaper at scale?

Sometimes at very high sustained RPS if you already operate staking ops - include engineer and hardware cost.

What is Yellowstone?

gRPC Geyser interface - Triton and others expose for indexers (Yellowstone gRPC).

Should CI use production RPC key?

Use restricted devnet key with low rate cap - never commit mainnet keys to public repos.

How do providers relate to Agave 4.1.1?

Providers run validator versions on their rollout schedule - occasional method/version skew - monitor status pages.

Can I use multiple providers?

Yes - primary/secondary failover or split reads vs writes if vendor supports.

Do providers cache getAccountInfo?

Some do briefly - do not rely on cache for post-transaction reads without refresh.

What about geographic latency?

Choose region closest to users and servers; test getSlot RTT from deployment region.

RPC Basics - JSON-RPC fundamentals
WebSocket Subscriptions - WSS endpoints
The DAS API - provider DAS features
RPC Best Practices - resilient client patterns

Stack versions: This page was written for Agave 4.1.1, Solana CLI 3.0.10, Anchor 0.32.1, anchor-lang 0.32.1, Rust 1.91.1, @solana/kit 7.0.0, Surfpool 0.12.0, and LiteSVM 0.6.x.

Recipe

Quick-reference recipe card - copy-paste ready.

import { createSolanaRpc } from "@solana/kit";
 
// Provider (API key in URL or header per vendor)
const rpc = createSolanaRpc(
  "https://mainnet.helius-rpc.com/?api-key=YOUR_KEY"
);
 
await rpc.getSlot().send();

# Self-hosted (operator)
agave-validator --ledger /data/ledger --rpc-port 8899

When to reach for this:

Mainnet or devnet apps exceeding public RPC rate limits.
Needing DAS, webhooks, or priority fee estimate APIs.
Running indexers requiring dedicated stake and bandwidth.
Compliance requirements for private RPC ingress.

Working Example

import { createSolanaRpc, createSolanaRpcSubscriptions } from "@solana/kit";
 
const HTTP = process.env.RPC_HTTP_URL!;
const WSS = process.env.RPC_WSS_URL!;
 
const rpc = createSolanaRpc(HTTP);
const rpcSubscriptions = createSolanaRpcSubscriptions(WSS);
 
const slot = await rpc.getSlot().send();
console.log("slot:", slot);
 
// Health check script for CI
if (slot < 1) throw new Error("RPC unhealthy");

What this demonstrates:

Separate HTTP and WSS URLs from provider dashboard.
Environment variables keep API keys out of source control.
Simple getSlot probe validates endpoint health.

Deep Dive

How It Works

Managed providers run Agave validators plus load balancers, caching, and indexer sidecars.
API keys authenticate and meter requests per plan tier.
Self-hosting runs agave-validator with your own stake, ops, and failover.
Feature matrix (DAS, gRPC, webhooks) varies by vendor - not commodity RPC.

Provider Comparison (high level)

Option	Strengths	Trade-offs
Helius	DAS, webhooks, dev tooling	Vendor coupling for advanced APIs
Triton	Yellowstone gRPC, performance	Ops integration complexity
QuickNode	Multi-chain familiarity	Solana-specific feature depth varies
Public RPC	Free, zero setup	Rate limits, no SLA
Self-host	Control, privacy	Staffing, stake, reliability burden

TypeScript Notes

// Failover: try secondary URL on 429/503
async function rpcWithFallback<T>(fn: (rpc: ReturnType<typeof createSolanaRpc>) => Promise<T>) {
  const primary = createSolanaRpc(process.env.RPC_PRIMARY!);
  try {
    return await fn(primary);
  } catch (e) {
    const secondary = createSolanaRpc(process.env.RPC_SECONDARY!);
    return await fn(secondary);
  }
}

Gotchas

Embedding API keys in frontend - keys get scraped from browsers. Fix: proxy reads through your backend for sensitive keys.
Single provider dependency - outage stops your app. Fix: secondary URL failover and cached critical reads.
Using free tier for production mint - 429 during spike. Fix: paid plan sized to peak RPS.
Mismatched HTTP/WSS hosts - subscriptions fail silently. Fix: copy paired endpoints from dashboard.
Self-host without monitoring - disk full on ledger stops RPC. Fix: ledger alerts and redundant nodes.

Alternatives

Alternative	Use When	Don't Use When
Public devnet RPC	Prototypes	Production mainnet
Colocated validator + provider hybrid	Trading firms	Simple NFT mint site
IPFS-only metadata	Static assets	RPC needs
Light client (future/experimental)	Extreme decentralization	Current production default

FAQs

Do I need a provider for devnet?

Public devnet works early; provider improves reliability for team CI and demos.

How do I pick plan size?

Estimate peak RPS from MAU, GPA weight, and websocket connections - load test before launch.

Can one key serve HTTP and WSS?

Usually same API key on paired URLs - verify vendor docs.

Is self-host cheaper at scale?

Sometimes at very high sustained RPS if you already operate staking ops - include engineer and hardware cost.

What is Yellowstone?

gRPC Geyser interface - Triton and others expose for indexers (Yellowstone gRPC).

Should CI use production RPC key?

Use restricted devnet key with low rate cap - never commit mainnet keys to public repos.

How do providers relate to Agave 4.1.1?

Providers run validator versions on their rollout schedule - occasional method/version skew - monitor status pages.

Can I use multiple providers?

Yes - primary/secondary failover or split reads vs writes if vendor supports.

Do providers cache getAccountInfo?

Some do briefly - do not rely on cache for post-transaction reads without refresh.

What about geographic latency?

Choose region closest to users and servers; test getSlot RTT from deployment region.

RPC Basics - JSON-RPC fundamentals
WebSocket Subscriptions - WSS endpoints
The DAS API - provider DAS features
RPC Best Practices - resilient client patterns

Stack versions: This page was written for Agave 4.1.1, Solana CLI 3.0.10, Anchor 0.32.1, anchor-lang 0.32.1, Rust 1.91.1, @solana/kit 7.0.0, Surfpool 0.12.0, and LiteSVM 0.6.x.

RPC Providers

Search Documentation

RPC Providers