Storage Layout Optimization

Apache Doris exposes the on-disk storage layout to users as tunable parameters. By configuring the storage layout properly, you can reduce storage cost, accelerate wide-table point queries, and lower the storage overhead of historical and cold data, all without changing the table model or queries.

Data Compression

Choose a compression algorithm per column (LZ4, ZSTD, Snappy, Zlib, and others) to flexibly trade off storage cost against query CPU and decompression speed.

Hybrid Row-Column Storage

Add a row-store copy on top of columnar storage to merge multi-column point lookups into a single IO, significantly improving high-concurrency primary-key point queries and wide-table SELECT * scenarios.

Hot-Cold Data Tiering

Move cold data down to low-cost media (HDD, object storage, HDFS) while keeping hot data on high-performance storage. Covers three modes: storage-compute separation, local tiering, and remote tiering.