November 22, 2005 at 5:04 am
Hi there,
I have this guy telling me, that a Data Transformation Task will perform better, if the picture shown in the transformationtab bundles all transformation lines in 1. When the transformationlines are shown as single, from column to column, this would degrade performance.
I find that hard to believe, but I don't know what happens under the hood. I would think when compiling a packages these transformations would be translated to an equivalent of INSERT INTO TABLE1(c1,c2,c3) SELECT c1,c2,c3 FROM TABLE2. I would be amazed to find out, that the picture presented would influence the statement generated.
But, as I said, I don't know what happens under the hood. Can anyone explain how it is? Is there a site somewhere, where I can find out about these things?
Greetz,
Hans Brouwer
November 22, 2005 at 5:50 am
This guy is telling you the truth.
Excerpt from microsoft site:
'N-to-N Column Mappings
Transformations using N-to-N column mappings require a matching number of multiple source and destination columns. You use this mapping in situations where each source column must have a corresponding destination column, and it is more efficient to configure all the transformations together (as a single data pump operation) rather than as separate transformations called individually for each row. By default, the Data Driven Query task uses this type of mapping configuration. Note - A single many-to-many Copy Column transformation is faster then many one-to-one Copy Column transformations. '
See this link: http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dtssql/dts_elemtr1_0feb.asp
November 22, 2005 at 6:07 am
Tnx Jonathan,
It still does not tell me why. But: I found that 1 too. It seems a COM is created for each mapping. Bundling the columns will create 1 COM, thus improving performance.
Greetz,
Hans Brouwer
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply