Planning Analytics

Hi Ian,

same migration context as the Metrics thread, so we looked at this as well.

There is no supported DEMANDLOAD equivalent in TM1 v12 / Planning Analytics as a Service.

In v11 this lived in }CubeProperties. In TM1 Database 12 that control cube is gone. Only a subset of cube properties moved to the REST API (VMM/VMT, calculation threshold, persistent holds, lock, measures/time dimension). DEMANDLOAD is not among them. CubeUnload is also discontinued.

So you cannot set “load this cube only when accessed” per cube the way you did on-prem. There is no documented v12/PAaaS replacement for that setting.

If the goal is still to reduce idle RAM, the practical v12 levers are different: model design, query memory guardrails, and the Metrics endpoint to see which cubes carry the footprint — not a demand-load flag.

If anyone from IBM has a supported replacement on 12.6.4, that would be useful to document. We have not found one.

Hi just having a look to see if you can do this via TI.

There are conflicting views.

Cubewise say you can.

https://cubewise.com/functions-library/tm1-function-for-ti-cubeunload/

IBM say it's deprecated for V12.

https://www.ibm.com/docs/en/planning-analytics/3.1.0?topic=12-deprecated-features-components-in-tm1

But from some of the demos i've been on things like CubeSetLogChanges have been sacrificed for the High Availability (HA) side.

So I bet its the same for unloading cubes as it would have to propagate cube updates live.

But this is just my opinion.

Thanks Julian — useful distinction.

Ian’s question is specifically DEMANDLOAD — i.e. not loading the cube when the database starts. That property did not move to the v12 REST cube properties, and there is no documented per-cube replacement.

CubeUnload is a different control: unloading an already loaded cube at runtime. Cubewise lists the function for versions 11 and 12, while IBM lists CubeUnload as discontinued in TM1 Database 12. I would treat the IBM documentation as the supported position for PAaaS 12.6.4, and the Cubewise entry as a function catalog that still mixes v11 and v12 items (CubeSetLogChanges / CubeSaveData are in the same list).

Even if CubeUnload still executed in some v12 environments, it would not restore DEMANDLOAD behaviour. Unused cubes would still load with the database; you would only try to drop them afterwards, and in an HA/Replica setup that is a different problem.

If someone has actually run CubeUnload successfully on PAaaS 12.6.4, that would be worth confirming separately. It still would not replace DEMANDLOAD.