Performance of View Zero Out

[Emixam]

Hello,
I started a project and one of the requirement was to be able to ViewZeroOut a very large data set (100,000,000+ cells). I never had to ViewZeroOut such a large data set and I noticed it takes around 10min.

I was wondering if there is a rule of thumb in term of number of cell per seconds that are VZO (ex: 50,000 cells per seconds) and what impacts the performance of ViewZeroOut. Any idea how to make it faster ?

I created a view using one of our generic TI with only level 0 and I skipped conso, rule and zeroes in the view creation.

Code: Select all

ViewZeroOut('Cube Test', 'ViewName');

The cube have no rule.

Thanks and have a good day

[Wim Gielis]

You should always create a view with only level 0 and skipped consolidations, rules, zeroes.

The process }bedrock.cube.data.clear supports parallel clears:
https://github.com/cubewise-code/bedroc ... .clear.pro

That would be my next attempt.

[John Hammond]

Could be cube logging is on.

Also Much quicker to delete view via supporting dimension element deletion and re-insertion but this is not always possible.

Otherwise Mr Geilis and the bedrock libraries is your best bet.