If your workload needs more requests than the public rate limits allow, you can run your own node. The stack is a pair of services, op-reth and op-node, and ships three profiles:
- full-snap-node (the recommended default);
- full-node;
- archive-node.
When to use:
- full-snap-node is the recommended default. It is the fastest to start, but it needs a trusted reth peer to snap-sync from.
- full-node keeps the same pruned state, for when you have no trusted peer. It re-executes the chain from Ethereum, so restore a snapshot to avoid a long initial sync.
- archive-node is for explorers, indexers, historical tracing, and eth_call at old blocks. It needs the most disk and RAM and always re-executes the chain from Ethereum.
You do not need to run your own node for normal use when you do not need unmetered RPC access beyond the public limits, or you do not want to read the chain independently.
An archive-node is required for historical tracing and eth_call at old blocks. Your node holds no Whitechain keys and forwards eth_sendRawTransaction to the public RPC.
Please note that the archive node is the heaviest to run, so we recommend using it only if you need historical tracing or eth_call at old blocks.