DSXchange

Rows per link in a BO job that completely pushes the work down to the database would be pretty meaningless. Look at the example you are giving for performance tuning in another post: Row Generator --> Neteeza There is 1 link and the number of rows across the link is 1. Depending upon the database st...

Oracle treats empty strings as null. http://docs.oracle.com/cd/B19306_01/server.102/b14200/sql_elements005.htm http://docs.oracle.com/cd/B28359_01/server.111/b28286/sql_elements005.htm You can work around this by providing a default value in the Oracle DDL. Understand that other loads where the colu...

ROWID is Oracle row identifier and is unique for every single row in your table. Without knowing more specifics of your job, there are only two reasons that I can think of that you would have duplicates: 1. As Ray said you have duplicates in your source table 2. You are not partitioning and/or sorti...

Have you tried enabling partitioned reads? If the table is a partitioned table you can have it read from each Oracle partition. If the table is not partitioned you can still read in parallel: 1. Enable partitioned reads = Yes 2. Partitioned reads method = Rowid round robin 3. table name for partitio...

How are you building your source sequential file? From your explanation it, I am assuming that you are building the file. Are you using DataStage to build it. If you are using DataStage to build the sequential file, have you considered using a data set instead of a sequential file?

You can't force datastage to push down when it doesn't want to do so. Your only option would be to manually manipulate the job if it didn't push as much logic into the database as you were hoping.

chulett wrote:... unless you specify the key(s) in your schema file.

Has there been a change to allow schema files to identify keys? We had this discussion last year:

viewtopic.php?t=147155&highlight=

If a key specification now exists it would definitely provide more flexibility.

With RCP the best way to update/insert is to write the sql statements in a file, then reference the file for your sql statements. Having DataStage generate your sql statement won't work because it doesn't know the key for your data.

How many bytes are in each row? A wide 10 billion row table is going to take quite some time to extract. A fairly skinnty 10 billion row table should be manageable. Have you tried to perform a parallel read? In the connector, enable parallel read and then set rowid hash and the table name equal to y...

Your sql can alias a column or a concatenation of columns as 'keycolumn' and then you can sort and partition based upon that column. It will also be helpful to have the key column extracted under its own column name so it can be used to insert into any target using drop unused coumn (which would dro...

The name field should be set to Unicode in Teradata and not Latin. And the SQL Server NLS is not ISO-8859-1, it is likely windows-1252. Windows 1252 was developed before the Latin-1 and is a superset of Latin-1. Once you have the column set to Unicode and the NLS MAP set to windows-1252, and then cl...

no, nothing different from the client as long as they are straight inserts.

Does the parent table allow null in the state column? If it does not allow null then you should not be attempting to insert a null into the child table else if allows a null and there is null present in the state table then you should not be getting this error.

Just happened to look out here and see this question this morning. You can have a DSN entry as simple as this work. Driver=/opt/IBM/InformationServer/Server/branded_odbc/lib/VMsqls24.so Description=DataDirect 6.0 SQL Server Native Wire Protocol Database= LogonID= Password= Address=ipaddr, portno Quo...

1. First deadlocking the database is more complicated than just running somethign on one node. At a mininmum you can decrease the array and transaction size to decrease the number of records effected by each node performing an upsert. The SQl Server table itself can be tuned by your DBA. 2. If you d...