db2admin wrote:
Hello,
I am wondering if someone have done comparison of loading data using
cursor VS. Pipe.
Please share your experience. which one is faster ?
By "loading data using a pipe", do you mean something like:
mkfifo apipe
db2 "export to apipe of del select * from table"
db2 "load from apipe of del replace into another_table"
If so, I can't imagine how this could be faster than load from cursor.
DB2 has to externalize the data it writes to the pipe, vs. doing
everything in memory inside the engine (with a cursor).
For example, timestamp takes only 10 bytes to store internally, but
when externalized it's 26 bytes. There's similar gains for all numeric
data types as well as date/time.
Ian Bjorhovde
