Tech Talks After Hours: OneStream Fitness Part 2

Ready to level up your OneStream Fitness routine? Roland Galli and Tom Linton continue to break down more performance moves that matter—DUCS, RemoveZeros/RemoveNoData, and “use filters, not loops”—so you can build leaner rules and faster consolidations.

Key Topics Covered

• Data Unit Calculation Sequence (DUCS): the order in which OneStream processes calculations during a consolidate or calculate — understanding this sequence is essential for ensuring calculations run in the correct order
• Eight rule files per cube: OneStream executes business rules 1 and 2, formula passes 1–4, then business rules 3 and 4, repeating as needed — order matters and must be considered when writing rules
• API data dot clear calculated data: using this method to improve rule performance by clearing only calculated data rather than processing the full dataset, reducing unnecessary computation
• Remove No Data: a key efficiency measure that strips cells with no data from the dataset before processing — recommended as a baseline best practice for nearly all rules
• Remove Zeros: an even stronger cleansing step that removes zero-value cells in addition to no data cells — appropriate when zeros are not meaningful to the calculation at hand
• OneStream Fitness analogy: a dataset with 56 million cells (28M zeros, 17M no data, and only a small fraction of actual data) is the equivalent of carrying 60 pounds of extra weight — it slows everything down unnecessarily
• Rule loops do's and don'ts: guidance on how to structure loops in rules to avoid redundant database calls, processing overhead, and performance degradation
• Emphasis on filters: using filters to narrow down the data set a rule operates on — the more precisely a rule targets only the cells it needs, the faster it runs
• Live use case: a side-by-side comparison of an unoptimized rule versus an optimized version incorporating the fitness techniques, demonstrating measurable performance improvement
• Real-world performance impact: the improved rule saved approximately five-sixths of total calculation time — in a one-hour consolidation, that translates to going from 60 minutes down to roughly 10 minutes
• Chasing milliseconds at scale: individual optimizations may seem small, but on large datasets with many cells and repeated calculations, the cumulative impact is significant
• Solution Exchange resources: standard report sets within each application provide data unit size statistics; the System Diagnostics solution exchange solution offers additional performance monitoring tools