Up To Something from Hau Ngo on Vimeo.
Here is a familiar phrase for anyone who has worked in SAP BW or BI production support environment: ‘Short dump in the Warehouse’
Checking transaction ST22, you should see another familiar screen telling you about a deadlock issue.
Personally, I find it hilarious that the system message tells you to try again and hope the outcome will be different.
There are a few ways to get around the deadlock issue and reload your data successfully. I like the two options below because that means less hoops to jump through (from a business approval standpoint).
Option 1: Delete the Index
In BI 7.0, the process chain creation wizard is smart enough to delete and generate an index when loading data into an infocube. In BW 3.x, this data target administration had to be specified.
I was performing the ETL manually, so forgot this step. In general, deleting the index of an infocube will speed up the loading process. (I can find the SAP Note if anyone is interested.)
Option 2: Change the Package Size
If you would prefer not to bother the Basis team to increase the global memory size, then changing the DTP package size is another quick remedy for dead lock issues.
Since dead lock situations arise when a number of processes encounter a circular waiting pattern (where A waits for B, B waits for C, and C waits for A), decreasing the number of parallel processes may be a remedy.
Here, I changed the default size of a data package from 20,000 kByte to 5,000 kByte. Also, I changed the number of data packages from a previous value of 10 to 5.
After deleting the index on my cube and changing the package setting, I restarted the load.
The cube was loaded with 3.4 million records in under 40 minutes. That’s a success in my book. =D
Posted via email from hau
A client
requests a specific report to be saved as a flat file.
In this case, the format
is a comma space vertical (CSV) delimited file, which is a common workaround
for Microsoft Excel's limitation of 65,000 rows. That sounds easy enough.
Working with SAP Application Process Designer (APD), however, was anything but straight-forward. The Business Explorer (BEx) query report was showing one set of numbers while the flat-file seemed to display whatever it wanted at the moment. After hours of banging my head against my desk, I have discovered 3 tips that will help to save you the frustration and recoup several hours from your workday. Here is a summary of the three tips followed by a more detailed explanation.
Three Rules for Working with SAP APD
1. Unhide all fields
2. 'Display Data'
3. Create formulas in APD
Before explaining the three rules in detail, I want to
show you a before picture. Below is a screen-capture, comparing the query
results with the generated flat file. The value of "50" for
Safety Stock in the report is correct but the value of "0.000" in
the APD flat file is wrong. (Click on the image below for a larger picture)
To drive the point home, here is another view of the same
record. Notice how the value of "50" for Safety Stock (in the
query) appears under "Reorder Point" (in the APD flat-file)?
Also, how the value of "24" for Open SO Quantity in the query
appears under Max Stock Value in the APD flat-file.
What is causing the values to jump around and be displayed in arbitrary columns? For that, we will need to take a closer look at the query definition.
Notice below how there are a series of green and blue icons
in the columns section of the query designer. The green icons are your
visible fields and the blue icons signify that a field (key figure,
calculation, formula, restricted key figure, etc.) is hidden from view.
Not only is that field hidden in the report output but it is hidden from the
APD mapping as well.
Rule 1: Unhide all fields in APD
Your life will be much better for it. See below.
Apparently, un-hiding the fields in Query Designer help APD to put the output values in the correct buckets. Otherwise, selectively displaying and hiding column fields will make numbers omelet look like scrambled eggs.
So now that you are displaying all of the report fields, how
do you go about verifying that the flat-file output will match the Business
Explorer (BEx) query? Well, you test the APD of course! How you go
about testing the APD is another thing altogether.
To start this topic, here is a sample APD. Notice there are
five distinct steps between the initial query result (step 1) to the final
flat-file layout (step 5).
So the question is "If the query result from step 1 does not match the data in the final flat-file in step 5, then how do you find out what went wrong and where it went wrong"?
One approach (that I am embarrassed to admit guilt of doing) is to run the ADP, compare the results from step 1 and step 5, and adjust the APD. That is, I tried adjusting the join in step 2. When that didn't work, I tried adjusting the calculation in step 3. When that didn't work, I tried adjusting the display/hide field mapping in step 4. When that didn't work, I would bang my head against the desk, rinse, and repeat. It's not the best approach or most methodical approach.
Rule 2: Display Data
A more pragmatic approach is to use the "Display
Data". See below:
Performing a Right-Click on a node (steps 1 through 5) will display the data at it exists at that point in the execution of the APD transformation.
For instance, I realized that my selective display and hide of the report fields was causing an issue after having verified the query results by selecting "Display Data" on the Query Result node at Step 1.
Rule 3: Create formulas in APD
After verifying the query output in the APD at step 1 matches my expectations, I select "Display Data" at the Join node in Step 2. At this stage, I find that my calculation for Open Purchase Order Quantity and Open Sales Order quantity was displaying the results in scientific notation. For example, the value "24" was displayed as "2.4000000000000000E+01". That was easily fixed by performing the calculation in the APD as a node, instead of the performing the calculation in the query.
Well, those are my 3 tips to making
working with APD a better experience. I hope the tips saved you a few hours. =D
During our last trip to Kaua'i, there were chickens at the Hilton. This time around on The Big Island, there are a number of cats walking around the Waikoloa Hilton.
They are cared for by a group of volunteers. Signs are posted, asking visitors not to feed the cats as it interferes with their diet.
Walking around the resort, I see a number of guests rubbing and petting the cats. They seem friendly enough.
I took this picture with my iPhone while waiting for a sunset. She was content to sleep under the bush, near my feet as I fiddle with my camera. She even purred when I reached down to scratch her neck. Melissa gave her a tummy rub.
She gave Melissa a small rash when she licked her hand. Melissa is allergic to cats, apparently.
Posted via email from hau
Funny how things work out.
I spent the better part of an hour trying to capture the sunset from our resort at the Waikoloa Hilton with my fancy digital SLR. However, the preview in the back of the camera just didn't captures that my eyes were seeing.
After the sunset, I took a picture with my iPhone to geo-tag the location. The above picture, taken with my humble phone camera, was closer to the real colors of the sunset. Funny.
Posted via email from hau
I think we're on the right plane. =P
Posted via email from hau
Melissa and I had a three-hour layover in Salt Lake City, Utah.
I had two coffees since 3:00 am so I thought taking some pictures would be cool. Of course, this could be the coffee's idea.
Here's a view from the D-gate.
Posted via email from hau
Recent Comments