Caching versus Intelligent Cube ?? 1

[mstrgrl]

If I want to run some type of report that will allow many data elements from my warehouse to be available for user reporting in a cache (so that when users request the elements on their reports they don't have to wait so long), would I want to do some type of report caching or intelligent cube?? I know a little about the differences, but I haven't really understood enough to compare the two yet.

Thanks!!!

 

[chael]

Intelligent cube basically is just a big cached report...and there is a size limit (in theory) to the cache/cube itself...I think it is around 5 mb but I am not sure and have never tested this.

So, if your report is going to be huge (1 million rows x 40 columns) you'll probably crush the cube -- only way to find out is to test it, really.

If your report is going to be of reasonable size, then go nuts. Hopefully your licensing, etc. is in compliance, and I think this is the type of reporting decision that needs to come on a report-by-report basis.

I think our friend entaroadun has some valuable input on what to do / not do with intelligent cubes...let's see what he says.

 

[entaroadun]

mstrgrl,

It's unclear to me exactly what you want to do. Intelligent cubes allow report reuse; basically, you can take a report, move objects off of the report template and into the report objects pane, and the report is automatically recalculated by the analytical engine. For example, if you have a report with Year, Quarter, and Month, and you take Month off of the report template and into the report objects pane, then the analytical engine will recalculate everything to go to the Quarter level.

This only works if you start with the Month report and retrofit it into the Quarter report. If you create a completely new report at the Quarter level, it won't use the cache for the Month report even if you have exactly the same objects on the report.

It sounds to me like all you want to do is increase report performance in general for a particular set of attributes or metrics. Intelligence cubing won't do it for you. You'll need agg tables in the WH.

 

[chael]

Of course the other option is to schedule reports, put them into the cache ahead of time, and voila, they are they when users need them.

This doesn't work for adhoc/prompted reports, though.

Give us some numbers to chew on as well as some more of the high level goals you are looking to achieve.

Chael