Summarizing SAP IQ’s Petascale Schema Model
Over the years, SAP IQ has continuously evolved its strategies for dealing with big data. Data size was traditionally limited by storage cost. Recently that became less of a limiting factor with multi-TB drives and SSDs. In such petabyte scale systems, IQ has a new schema model to keep the database lightweight and performant. To summarize, fact tables are hash partitioned on either selection or join columns and indexes are only heavily used on dimension tables. Fact tables use hash filtering if the table is partitioned on the selection column. The remaining keys are eliminated using N-bit scans or index lookups. The net result is fast and efficient processing with minimal storage size. We envision this as a key aspect of achieving petascale performance. For best results, use the latest version of IQ.