Skip to content
Saiku Cloud

Analyze

Analyze is where you go to actually ask a question. It shows every cube in your tenant as a card, tells you whether each one is healthy, and puts you one click from querying it.

Once your cubes are set up, this is the page people open most. Analysts come here for ad-hoc questions; everyone else uses the health badges as a sanity check when a number looks wrong.

Opening a cube

Click Open in Saiku on a healthy cube and you land in the Saiku UI with the cube loaded — same identity as the dashboard, no second login. Drag dimensions onto rows or columns, drag measures into the middle, and the pivot rebuilds as you go. Save the workbook from the toolbar if you want to come back to it.

The Saiku UI has the full OLAP toolkit: drill-down through hierarchies, filters, sorts, totals and sub-totals, export to Excel / CSV / PDF, and an MDX editor for people who’d rather write the query than drag it. There’s a proper guide to the workbench, but three details save the most time:

  • Visual totals. The totals toggle has a visual totals mode that makes a parent row sum only the children you kept after filtering, rather than the whole cube. Reach for it whenever a filtered pivot’s totals look implausibly large.
  • Hidden measures. Measures marked invisible in the schema don’t appear in the picker. Admins get a show hidden measures toggle when debugging a cube.
  • Session expiry. If your session times out mid-analysis you get a banner, not a blocking dialog. Your pivot stays on screen untouched and one click signs you back in.

Cube health

Filter chips across the top count cubes by status, so you can jump straight to whatever needs attention:

  • Active — loaded cleanly, ready to query.
  • Broken — the engine couldn’t attach the cube, and the launch button is disabled. Hover it for the actual error. Almost always one of two causes: the connection is misconfigured, or the schema references a column the warehouse doesn’t have. Fix it at source and the status updates within a few seconds.
  • Unconfigured — the schema was uploaded without a warehouse bound to it. The card offers Bind a connection instead of a launch button.
  • Loading — the engine is warming the cube. Common on a cold start or a slow warehouse; the page keeps polling until it settles, so you don’t have to refresh.
  • Unknown — we couldn’t determine the state. Usually transient; reload before worrying.

Each card also shows what’s inside — cube, measure and dimension counts — which is the quickest way to spot a schema that parsed but came back emptier than you expected.

The Saiku UI vs. the API

The Saiku UI is for humans: interactive, point-and-click, the right tool for an analyst chasing a hunch. For anything machine-driven — an agent, a custom dashboard, a scheduled report — use the API instead. Both hit the same engine and read the same cubes, so it’s purely a question of which front end fits the job.