Anyone who works with BPC will no doubt be aware of the Lite Optimise process. It's easy to trigger, runs relatively quickly and significantly improves performance. Essentially, a Lite Optimise roll's up and compresses data in the BPC model / cube then updates the DB Statistics.
However, if you're new to BPC, there are a few other things you need to be aware of in order to help you understand the process a little more. There may be more things to add to my list so feel free to comment on the post!
Be aware if people are trying to save !
If you were to take a closer look at the steps involved in a lite optimise via the process chain (/CPMB/LIGHT_OPTIMIZE) you will see a couple of important steps:
Notice the highlighted steps above, in order for the roll-up and compression to work, the cube needs to be taken out of planning mode. It is this planning mode which allows your Input Schedules to write updates back to the cube in real-time.
So, it's very important to understand that if the Lite Optimize process is running and the cube is not in planning mode the user will not be able to save data !!
Currently, BPC doesn't do a very good job of letting you know why you can't save (e.g., because a Lite Optimize is running) so be sure to trigger it when saving activity is at it's lowest!
Big data loads require a Lite Optimise
Any large data changes to a model need to be followed by a Lite Optimize. This could include data loads from BW, clear and copy packages or substantial user submissions via Input Schedules. You will begin to notice deterioration in performance the longer you go without a Lite Optimize!
In our environment, we have tagged a trigger for the Lite Optimize to kick off after every transactional load process into BPC from BW. Admin users are also advised to kick off a Lite Optimize after running copy/clear/load operations The next point explains how this can be done.
Trigerring a Lite Optimize outside of the EPM add-in
It is true, the lite optimize can be manually triggered or scheduled from the EPM add-in. But what if you have a separate BPC process chain and want to automatically trigger a lite optimise once it is complete ? (e.g., daily data loads from BW into BPC)
You can do this as follows:
1. Create a new process chain with a start process
2. Create a new ABAP Program
3. Set the 'Program Name' to 'UJD_TEST_PACKAGE'
4. Create a new 'Program Variant'
5. Fill in the details (e.g., Package ID = Lite Optimize, Package Group ID = Other Function (or whatever folder the Lite Optimize package is stored within EPM)
6. Reference the newly created chain at the end of your other process chains.
If you do a search on UJD_TEST_PACKAGE you will find a lot more detail on these steps which all also allow you to trigger any package from BW !
Get rid of the 0's!
Whenever you run a clear package, you're not actually deleting data from the cube. Infact, you are settings key figures to 0. This would explain why you cannot delete a member from a dimension after 'clearing' all data. There is still data against that member, albeit zero's!
A great way to clear these 0's is to update your Lite Optimize process chain to clear 0's as part of the Collapse step in the chain /CPMB/LIGHT_OPTIMIZE.
You should really avoid using the selective deletion function on the cube! One slip of the hand or an incorrect selection of data and you're in a spot of bother.
Lite Optimise before you back up an environment!
UJBR is a great BW program which allows you to backup and restore your BPC environment. If you don't know about this, you should certainly do some searches to bring yourself up to speed! We run this every night as an automated job to ensure we have a rolling 7 day backup of our BPC environment which we can easily roll back to without impacting the rest of BW.
Anyway, it can take quite a long time to backup your environment if you have a lot of data. Using a Lite Optimize immediately before a backup will significantly improve your wait times !
That's all that comes to mind right now. I hope this helps those of you who are starting out in BPC. I certainly could have done with these tips early on !
As already mentioned, if you can add to this in anyway, please comment below.