Skip to content

Scan Mode

Discover candidate cubes in an instance, ranked by RAM consumption. Use this as a triage step before deciding which cubes are worth optimizing.

When to use

  • You're starting fresh on a model and don't know which cubes matter most.
  • You want to batch-generate per-cube JSON configs for the cubes worth benchmarking.
  • You need a quick inventory of which cubes have a custom dimension order vs the default.

CLI

optimuspy scan --instance tm1srv01

--instance is required and must match a section in config/config.ini. By default the list stops at the cubes that together account for 60% of the model's RAM; --ram-percent 80 widens it (the cube that crosses the threshold is included).

Output format

A console table sorted by RAM, descending:

Cubes accounting for up to 60% of total model RAM (12.43 GB), not yet optimized:

   #  Cube Name           Dims  RAM (GB)  % of Total  Dimension Order
   ─  ─────────────       ────  ────────  ──────────  ────────────────
   1  Sales                  8      4.21       33.9%  ['Time', 'Version', ...]
   2  Budget                 6      2.84       22.8%  ['Year', 'Department', ...]
   3  Forecast               7      0.95        7.6%  ['Time', 'Scenario', ...]

  Total: 3 cubes, 8.00 GB (64.4% of model RAM)

Already-optimized cubes (where the visible dimension order differs from the storage order) are always excluded on the command line, because a cube someone has already reordered deserves a look before it's benchmarked again. To see them, use the UI's Optimize page with Include optimized ticked.

Generating config files

Add --output with a directory to write one JSON config per candidate, ready to feed back into optimuspy optimize:

optimuspy scan --instance tm1srv01 --output cube-configs/auto/

Each generated file uses safe defaults (executions: 5, output: csv, no views/processes). Edit each one to add views, processes, or position rules before running.

Use from the UI

The Optimize page runs the same scan automatically when you connect to an instance, with a slider for the RAM threshold. The CLI mode is mostly for scripting (e.g., a nightly inventory of cubes that need attention).