APAR status
Closed as program error.
Error description
With BI 10.2.1, the query engine is optimized to work with symmetric queries only where asymmetric queries can potentially have poor performance compared to previous releases. This issue has been observed only against TM1 datasources in DQM mode.??An example of a symmetric query would be a report that only contains nested members on an edge with no siblings, such as:? Column#1: <#Sales Price Sum#>?ROW#1: <#children(CANADA)#> <#children(2010)#> <#1234#>?ROW#2: <#children(2010)#> <#1234#>?ROW#3: <#children(CANADA)#> <#children(2010)#> <#1234#>?ROW#2: <#children(2010)#> <#1234#>??It's symmetric in that every member in the query intersects with each other.??Asymmetric queries can occur when sibling sets are introduced on the edge, such as:?? Column#1: <#Sales Price Sum#>?ROW#1: <#children(CANADA)#> <#children(2010)#> <#1234#>?ROW#2: <#children(2010)#> <#1234#>?ROW#3: <#children(JAPAN)#> <#children(2011)#> <#1234#>?ROW#2: <#children(2011)#> <#1234#>??Where not every member in the query intersects with each other. In this case, the asymmetric part is from the children of Canada intersecting with the children of 2010, whereas the children of Japan intersect with different members from 2011. So Canada does not intersect with 2011 and Japan does not intersect with 2010 in the report.??The reason this is a problem is BI will query TM1 for the data with a single simplified symmetric query. So even though the asymmetric report does not request Japan and 2010 intersections, the query we to TM1 will ask for those intersections. This is added overhead and extra calculations are done on TM1.??We are working on a future release of BI that will work with asymmetric queries more efficiently so extra data is not computed in this way.?
Local fix
These are only potential workarounds. They may not be applicable to every report where asymmetric queries are used. Workaround #1: When there are rows in the crosstab report that do not have a lo t in common with each other and thus constitute an asymmetric qu ery, the report can be split up into multiple different crosstab s, one for each row in the original report. This way each query will be symmetric and no extra data will be calculated in TM1. Workaround #2: The second option is to wrap a member() function around the memb er in the first row. This will force the engine to handle the t hat member row independently from the rest of the edge as though it were in it's own query. This leaves the subsequent rows as being asymmetric, but there are not many additional intersection s when combining these into a single query so there is less of a performance impact. Most of the performance hit comes from the first row. This is slightly slower than the first workaround, but in this case the report output will be the same as the origi nal. Please note this workaround is only temporary for this release o f BI. This workaround will also cause less memory to be used in BI and allow the query to run in seconds once TM1 has calculate d the values from a cube update. The first time the query is ru n after a cube update should be closer to the previous release o f BI.
Problem summary
**************************************************************** * USERS AFFECTED: * * All Users * **************************************************************** * PROBLEM DESCRIPTION: * * See error description. * **************************************************************** * RECOMMENDATION: * * Upgrade to IBM Cognos Business Intelligence 10.2.1 Fix Pack * * 2 * ****************************************************************
Problem conclusion
Code Fix
Temporary fix
Comments
APAR Information
APAR number
PM98021
Reported component name
COG ADMINISTRAT
Reported component ID
5724W12AD
Reported release
A21
Status
CLOSED PER
PE
NoPE
HIPER
NoHIPER
Special Attention
NoSpecatt
Submitted date
2013-09-27
Closed date
2014-01-31
Last modified date
2014-01-31
APAR is sysrouted FROM one or more of the following:
APAR is sysrouted TO one or more of the following:
Fix information
Fixed component name
COG ADMINISTRAT
Fixed component ID
5724W12AD
Applicable component levels
RA21 PSN
UP
[{"Business Unit":{"code":"BU048","label":"IBM Software"},"Product":{"code":"SSEP7J","label":"Cognos Business Intelligence"},"Component":"","ARM Category":[],"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"10.2.1","Edition":"","Line of Business":{"code":"","label":""}}]
Document Information
Modified date:
06 March 2023