Question:
I have a very large volume of data that I have pulled from PA about 7.5 Million rows into an ODS in BW. What is the best way to pull this data forwards since I am finding that I am constantly getting dumps and not being able to load downstream ODS objects and cubes?
SHould I pull a smaller volume first and then do a delta from that point?
How do I do that? How do I avoid inconsistencies?
Thanks a lot
Answer:
Could you please tell us which dumps you get?
MAybe you could do more than just one full load (use selctions!!) because I think delta load are less prformant than full loads....
You could go on like this:
full load 1 0fiscper 001.2002-001.2002
full load 2 0fiscper 002.2002-002.2002
(and so on, there you should find a compromise between the amount of load needed and the data pulled per load), 1 million records per load shouldn't make problems (this work in my test system even).
In addition to that you shouldn't load and activate the data in one step. Also the load is better if the ODS isn't enabled for reporting (no SIDs have to be generated then)
_________________
Juergen Limbach
Answer:
Juergen, the data is already in an ODS, the problem is between the ODS and a further datatarget, so not at extraction time.
between ODS and datatarget and cannot mix full loads and delta load, so -as long as I know- your only solution is the one you tried to use.
As Juergen stated, please be more explicit about the type of shortdump you receive ; perhaps is it only a timeout, in that case it is easy to handle.
Also, I don't think 7.5Millions records is that huge, especially for a COPA ods.
Ch
_________________
_
There are only 10 types of people in the world :
those who understand binary and those who don't.
Answer:
Not my day today....
Sorry
Isn't it possible to define the selection criterias within the export datasource / infosource / infopackage. I did that and then it should also be possible to load all data via 1-n fulloads and afterwards do a delta initialization without loading data?
_________________
Juergen Limbach
Answer:
ok we extract data into an ODS and we have no problems with this at the moment. But in the future if we are forced to re-initialize, the data will be a lot bigger and I am not sure that even after changing the parameters for data loads etc, that we will be able to load data ina sane amount of time.
1. Assuming I have a very large COPA, in excess of 2 million rows in R/3 what is the best way to get this data into BW into an ODS.
2. In regards to the ODS dumps, it is related to note: 578150 as far as I can tell. We have aplied this note and we are currently testing. Will let you know. If in the meantime if someone can shed some light on issue 1. that would be great
Thanks a lot
Answer:
Then first answer of JuergenL apply for your issue 1
Ch
_________________
_
There are only 10 types of people in the world :
those who understand binary and those who don't.
Answer:
There isn't any export datasource here. just one ODS feeding the next. How can I define the selection criteria if any?
If I follow the first suggestion then when i am on July 1 I do a full load and then from then on, I delete the previous load of July 1 and do another full load for july2.3.4...... until the end of the month
any isssues with data consistency?
Answer:
The one ODS feeding the next IS an export data source, in the AWB - infosources hit strg f3 and you will see something like 8ZZYOURODS there you can maintain the selections in your info package (I guess you don't have to change something within the datasource via rsa6)
_________________
Juergen Limbach
Answer:
you guys are right...
the selction criteria can be maintained there as you guys say.
One question though, if we do do 1-n full loads, and if after that i do a delta init, will it cause any inconsistencies?
Answer:
I Suppose the dollowing:
1-n full loads until the whole data is loaded. Then a delta initialization without data load (there's a checkbox in the info package) and then normal delta loads.
You have to make sure that absolutely NO data is loaded into the source ODS within the whole process of full loads and delta initialization. If that is granted I don't think there will be inconsistencies.
_________________
Juergen Limbach
Answer:
I see all the datasources : 8* in my development system, but i DO NOT SEE THEM in my Quality system. What gives? I can do deltas from one ODS to the next and i know they are there, but how can i see them? I tried transporting them but they do not show......
Answer:
/forums/viewtopic.php?t=52843&highlight=generated+objects
_________________
_
There are only 10 types of people in the world :
those who understand binary and those who don't.
Answer:
hey thanks, the problem now is that i have created the separate datapackages for each month, but i cannot do any loads, Each time, it fails immediately and says that i need to replicate the datasource,
Errors in the source system.....I have not made any chanegs to the datasources....
Answer:
Try to generate the 8* again, sometimes this is neccessary even if you don't know why...
_________________
Juergen Limbach