Partition Pruning
Partition pruning is a query optimization where the engine skips entire partitions of data that cannot match the query, so it only reads the partitions that matter. For time-partitioned data, this is one of the biggest speedups available.
How pruning cuts work
When data is partitioned, often by time, each partition holds a known slice, like one day or one hour. If a query asks for last Tuesday, the engine can skip every partition outside that range entirely. It never reads them, which saves enormous amounts of work on large datasets.
This is why partitioning strategy matters so much for time-series data. Partition by time, and time-range queries, which are almost all time-series queries, become dramatically cheaper because most partitions get pruned away.
Pruning pairs naturally with predicate pushdown, which skips data at an even finer grain within the partitions that remain.
How Arc handles Partition Pruning
Arc partitions time-series data by time, so the typical time-range query prunes away most partitions and reads only the relevant slice. That is a large part of why queries stay fast even over very large histories.
Arc is a high-performance columnar database. Open Parquet on storage you own, single Go binary, production-ready in 30 seconds.