Archive nodes, explained: what you actually pay for
The word "archive" gets used loosely across RPC providers, and the difference between a real archive node and a full node dressed up as one is the difference between a query that returns and a query that 404s. Here is what archive actually means at the storage layer, and why it matters for the queries you run.
Full node vs. archive node
A full node validates every block and keeps the current state trie plus roughly the last 128 blocks of historical state. It can tell you an account's balance now, but ask for its balance at block 10,000,000 and it recomputes nothing — the intermediate state was pruned. An archive node keeps every historical state root, so it can answer state queries at any past block directly. That completeness is the entire value proposition, and the entire cost.
Why the disk bill dominates
Ethereum archive state is multiple terabytes and grows continuously; the same is true in proportion for high-throughput chains like BNB Smart Chain and Polygon. Archive nodes live on NVMe because random-read latency into a multi-terabyte trie destroys throughput on spinning disks. That NVMe footprint — not CPU, not bandwidth — is why archive access is priced above vanilla RPC almost everywhere.
The queries that require it
Any eth_call or eth_getBalance with a historical block tag, trace_block and debug_traceTransaction over old transactions, token-balance reconstruction for accounting, and most analytics backfills all need archive state. If your product computes historical positions, reconciles ledgers, or replays transactions, a full node will silently fail on exactly the requests that matter.
How to verify a provider actually serves it
Do not trust the label — test it. Pick a contract call, run it with a block tag a few million blocks in the past, and confirm you get a result rather than a "missing trie node" or "state not available" error. Run debug_traceTransaction against an old hash. If either fails, you are on a full node regardless of what the pricing page says.
Sıkça sorulanlar
Do I always need an archive node?
No. If you only ever read current state and recent blocks, a full node is cheaper and sufficient. You need archive the moment you query historical state, trace old transactions, or backfill analytics.
Is archive data different from indexed data?
Yes. Archive nodes serve raw historical state via RPC; indexers derive higher-level datasets (balances over time, token transfers) on top of that. Reliable indexing depends on an archive source underneath.
Related
Bakiye yükle, anahtarını al, yayına çık.
Self servis. Kripto ya da kartla öde. Krediyle ölçümlenir — ağır ilkel yapılar daha pahalı, basitler ucuz.
API anahtarı al