Query Locality Split

Query Locality Split is the practice of separating the storage layer from the query layer in a client warehouse design, then matching each layer to the cheapest service that meets its actual access pattern. Most agency reporting workloads are lopsided: a small share of tables gets hit every dashboard refresh, while the bulk of client history is read a few times a quarter. Putting both on one platform means paying analytical-database rates for archival rows. Splitting them means a columnar engine such as ClickHouse serves the hot path at millisecond latency while object storage holds the cold floor. The trade-off is a second system to operate and a sync job to monitor, which is where scoping discipline matters. A concrete case: a client with three years of campaign history can keep the trailing 90 days in the query layer and push the rest to flat-rate object storage, cutting the bill without touching dashboard load times.