on 2008 Feb 04 1:37 PM
Hi,
We have legal Consolidation in place on BCS 6.0.
The performance of the query has been deteriorating. The 2007 consolidation is completed and now I would like to compress the cube to improve the performance.
Can you please is there any adverse impact other that request id is not available for deletion? Is there any impact on virtual cube?
Please advise.
Also inform the steps that are required to compress the cube.
Thanks in Advance,
Regards,
UR
Ur,
I think we were in the situation that you are in right now, and here is what he wad/did:
Our consolidations environment was setup with 5 hierarchies, 4 of which contained between 3.5K - 4.5K Profit Centers. Using the regular Virtual Cube, we had adequate performance for the P/L, but when we ran the Balance Sheet, the performance was terrible (25-45 mins for a balance sheet). At that time, we did compress the cube after each period was closed - The performance did increase, but the overall performance was still unacceptable.
As a result, we went to a new design (not the latest though), utilizing the request ID and the delta process. In this configuration, an additional cube was added - a reporting cube (CB14). The reporting cube and the virtual cube were placed in a MultiProvider and a delta process was setup: any open requests that were in C11 were closed and moved over to CB14. Using the defined exit variables, we get the closed records out of the CB14 cube and only the open ones from C11.
In the new setup, we cannot compress the cube, but I don't foresee any benefits from doing so.
In summary, compression has been encouraged by the SAP consultants that were assigned to this project. It did improve the performance.
Check the indexes on C11 as well - I had degradation of performance at one point (we generate about 220K records in BCS a month) and had IT take a look at it. After rebuilding the indexes, I returned to my normal "times". I had them enable the setting on the cube to automatically rebuild indexes after data update.
If you can, test this in your Dev environment first
Good luck
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Guys,
Thanks for valuable Input.
As per suggestion, I understand that the long term solution is to go for Data Mart/Mulitprovider Solution in line with white paper on ' How to ...configure the Delta Load based MultiProvider Solution'
However, I have very limited time and my first quarter (Nov-Jan) consolidation is due to start from 20th Feb. Under the circumstances, if I only compress the data for 0BCS_C11, will my reporting performance improve?
Please advise.
Regards,
UR
Hi,
I have been part of BCS implementation where performance was core issue with more than 500 company codes:
Following are my experiences to share:
1. Dan suggestion is good and that is first important step to do. Data mart the BCS data to another basic cube.
2. We have gone for partition of cube by period ,year. Please refer the following link for self help.. http://help.sap.com/saphelp_nw04/helpdata/en/33/dc2038aa3bcd23e10000009b38f8cf/content.htm
3. We have used aggregates in the cube to enhance the performance.. Refer the following link for some more details https://www.sdn.sap.com/irj/sdn/go/portal/prtroot/docs/library/uuid/e7deb490-0201-0010-f882-
4. More the navigational attribute enabling in the cube.. more impact on performance.. Keep it to minimum from reporting point of view
5. Try to keep the text data elements to minimum in the cube..
There are many more points to share. Incidentally we have to implement multiple step to achieve this since we had mega million records in BCS cube in the first year itself..is it your situation also
I have not seen the compression in BCS transaction cube.. instead you data mart the data as per step one as above to another BW basic cube and then compress the data and do the BCS reporting out f that
Compression of the cube request can be easily handled thru process chain(tcode: RSPC). You are doing good with right approach
Above all.. it is not long effort .. you are dong well with correct first thought.. to compress the data...
Warm regards,
Mani.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Mani,
Thansk for your input.
Well we were facing two major problems:
1. The Workbook/query on 0BCS_C11 Virtual cube is very low. The expansion and collapse of the hierarchy are taking alomost 60 seconds
2. Another problem was executing UCMON-Data Collection task. When data was being extracted from Staging Cube (this is getting data from DS 0FI_GL_4) to 0BCS_C11, it was taking lot of time for big companies. We are trying to reduce the data volume of BCS Staging cube by removing characteristics like Accounting Document Number.
Now we have closed financial year 2007. Do you think the query will still work slow as mentioned in point 1 even for the first quarter. In the quarter we are singe entry mandatroy variables for Fiscal Year.
Regards,
UR
UR,
1)When you include the Hierarchy in the Free Characteristics the data is brought in only when the user Navigates to the specific Hierarchy Node. Ideally it should be done the way you moved into Free Characteristics.
Review the following link for performance improvement guidelines:
2. Also try to do minimize data transformation in BCS data stream and instead to the same in Basic cube. This way you can enhance the performance of data stream.
It will be a good suggestion to post this in the BI forum so that you can optimize the opinion from many people .....
Good luck..
Mani
Edited by: Mani on Feb 7, 2008 1:51 AM
I'm sorry I do not know about the compression.
But, if the reporting performance is slow, SAP offers a white paper for using a datamart with BCS. It is offered for improved reporting performance.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi,
The best place to ask such a question is one of the BI forums:
/community [original link is broken]
You may get an answer in a matter of minutes. Try...
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
User | Count |
---|---|
8 | |
5 | |
2 | |
2 | |
1 | |
1 | |
1 | |
1 | |
1 | |
1 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.