Who can explain TPT instance count and its details

Tools
N/A

Who can explain TPT instance count and its details

In TPT apply clause, you can specify number of instances used for each operator. For example

APPLY ('INSERT INTO TARGET TABLE (COL1, COL2) VALUES (:COL1, :COL2);') IGNORE DUPLICATE ROWS TO OPERATOR ( LOAD_OPERATOR[2])
SELECT * FROM OPERATOR(READ_OPERATOR[3]);

In this sample, my load_operator is a update operator, while read_operator is DATACONNECTOR PRODUCER hook up to a single file.

specifying the instance on the loader and reader makes me wondering what will happen in reality:

1. does that mean the the file reader will have multiple file reading process reading different portions of file in parallel to boost reading throughput or the file reader is still one instance and the reader process file in blocked fasion first. Then the buffer is dispatcehd to different threads to process data (match schema, split fields etc)?
2. Load operator is the one really interesting here. UPDATE operator is equivalent to MLOAD. specifying multiple instance of UPDATE operator makes me wondering what is happening here. It is unlikely to have multiple MLOAD jobs insert into the same target table. Then how does the 2 load operator works? does that mean that it is still one mload job, but by specifying 2 instance, number of sessions will be doubled? If not so, how does the multiple instance of load operator helps the load performance?
1 REPLY
N/A

Re: Who can explain TPT instance count and its details

for those who are intersted to find the answer, please refer to here:
http://developer.teradata.com/tools/articles/use-tpt-to-move-data-without-landing-the-data-to-disk#comment-14221