DSXchange

The table is partitioned based on date and the pull needs to be performed for a particular date. This means my pull has all the data in one partition. so the partitioning in stage will not help.
mathewb

Here is the problem If I runthe select statement, the database is responding only at a speed of 5000records/sec since we have to go over dblink. Which means I can get max of 18million an hour( if I have only one child pulling the data). I sure PX will be able to do the sorting of these records in me...

Hi Ray I was not aware of the programming chapter of the documentation. I will go through that but I was searching for some samples which you guys have done..real time scenarios...Any code which is available in this side which some one has posted earlier. Regarding the Learning center, how soon are ...

I am looking from all the custom libraries which is currently available. I need to get an idea how the tool can be enhanced and how much people have achieved in this area. i am not able to find any place in this site where I could refer to libraries and custom codes. Is all this a part of future enh...

Ray, In any case..the write into file has to wait for all the records to be sorted in the memory. So the process will be slower than my current design unless the sorting in the memory is real fast. Also the process will exaust the server memory since it requires more than 5GB to hold everything in m...

No the table is not indexed. Since the pull from this table is over dblink, it cannot be more than 4000 records/sec. Since all the data resides in one partition of the table, I cannot use any datastage parallelism. I am currently making the 4 parallel pull with ranges (individual stages funneled to ...

Hi, Searched complete forum and found simlar topics but mine is little different. This is regarding sorting and would like to hear you guys opinion. I need to pull about 100 million records from Oracle over dblink and create a file sorted based on one column. All data resides in one partition. Curre...