eth_getLogs becomes increasingly expensive as chain history grows, and client-side indexing is one way execution clients can reduce the amount of history scanned for log queries. Geth already ships FilterMaps, a 2-D probabilistic log index, but reproducing it in another client requires matching much more than just the high-level data structure.

This presentation will cover the design of bringing that approach to Reth: matching Geth's global log index, RocksDB store, building the index incrementally from finalized blocks through a background service, keeping a bounded history window, and integrating it with the existing RPC path while falling back to bloom scans for unindexed ranges.