Individual committing for a master process

[macsir]

I know a master process will only commit data after executing the last sub process if it has multiple executeprocess functions. But can we have individual committing for it like single commit mode in chore?

[Darkhorse]

Savecubedata(cubea)

Turn off logging function in the TI it should commit after every TI that's written in

[macsir]

I think you mean CubeSaveData? Have you tried it before? I wondering how long it will take if I save a large cube.

[BariAbdul]

Savecubedata(cubea)

Turn off logging function in the TI it should commit after every TI that's written in

Darkhorse could you please elaborate further how turning off logging in TI would make individual process to commit in single commit mode,as far as i know disabling logging used for limit the log file grow too big.Thanks and appreciate your reply.

[TableManagerOne]

There may be some confusion in terminology - committing cube data to disk vs committing a transaction. BariAbdul seems to be talking about the latter, while Darkhorse may be talking about the former. Oddly enough, CubeSaveData may (unintentionally) provide the solution to both.

The 10.1.1 Fixpack Releases topic links to a tech doc, Locking and Transaction behavior of the SaveDataAll and CubeSaveData TurboIntegrator functions has changed which mentions:

Similarly, CubeSaveData will now begin and end with a transaction delimiting commit operation.

So, it seems one could conclude each child process with a call to CubeSaveData on a small and/or never changing dummy cube (to make the CubeSaveData operation quick,) which would have the effect of delimiting each child process with a transaction ending commit (similar to the multi-commit mode in chore.)

I doubt that IBM intended for CubeSaveData to be used in this way. Consequently, performance probably isn't optimal, and there maybe undesirable consequences. Also, it wouldn't be guaranteed to work in the future.