Writing about running RPC in production.
How providers meter usage, how to handle rate limits without losing transactions, and when a free public endpoint stops being good enough.
Querying Bittensor without running your own node
A Bittensor subnet miner or validator needs a reliable chain endpoint long before it needs its own archive node. This walks through what a managed endpoint gives you, which Substrate methods matter, and the failure modes that actually cost you a weight-set window.
bittensorinfrastructure
Why eth_getLogs is the call that breaks your RPC budget
One method accounts for a disproportionate share of most teams RPC spend and most of their timeouts. This explains what the node is actually doing during an eth_getLogs call, why block range is the variable that matters, and the query patterns that keep it cheap.
performancepricing
Migrating off a self-hosted node without a flag day
Moving from your own node to a managed endpoint does not need a cutover window. This is the incremental path: shadow traffic first, move reads, keep a fallback, and the specific things that break if you skip a step.
infrastructureoperations
JSON-RPC 2.0 error codes, and the ones a gateway adds
A reference for the error codes you will actually see from an RPC endpoint: the standard JSON-RPC 2.0 set, the node-level codes layered on top, and the gateway codes that tell you a limit was enforced before your request ever reached a node.
referencejson-rpc
Compute units, credits or requests: how RPC providers actually bill you
RPC providers meter in three incompatible units — compute units, credits and plain requests. The unit decides whether a trace-heavy workload costs the same as a read-heavy one. This post explains each model and how to compare quotes that are not denominated the same way.
pricinginfrastructure
How to handle RPC rate limits without dropping user transactions
Every managed RPC provider throttles you eventually. The difference between a brief slowdown and a failed transaction is what your client does with an HTTP 429. This post covers Retry-After, exponential backoff with jitter, request batching and the idempotency trap around sending raw transactions.
engineeringreliability
Public RPC endpoints are free. Here is what they cost you.
Every chain publishes a free public RPC endpoint, and they are genuinely useful for local development. In production they introduce aggressive shared rate limits, no archive depth, no analytics and no accountability when they go down. This post covers when the free endpoint is fine and when it stops being fine.
infrastructuregetting-started