Footprint Reduction And Uniqueness Enforcement With Hash Indices In Sap Hana

Databases commonly use multi-column indices for composite keys that concatenate attribute values for fast entity retrieval. For real-world applications, such concatenated composite keys contribute significantly to the overall space consumption, which is particularly expensive for main memory-resident databases. We present an integer-based hash representation of the actual values for the purpose of reducing the overall memory footprint of a system while maintaining the level of performance. We analyzed the performance impact as well as the memory footprint reduction of hash-based indices in SAP HANA in a real-world enterprise database setting. For a live production SAP ERP system, the introduction of hash-based primary key indices alone reduces the entire memory footprint by 10% with comparable performance.