Discussion:
DB2 LOAD Question/issue
(too old to reply)
t***@gmail.com
2009-06-10 20:18:10 UTC
Permalink
i'm loading data at maybe 250,000 lines per file using IMPORT and it's taking FOREVER.. i realize LOAD is supposed to be faster. but it doesn't seem to preserve the order of my source data.

i did a test where i LOADed 2 files (test1, test2), each containing 10 lines into a table.. they both worked perfectly (preserved data order, primary key generated correctly).. then i added another 200 records to test2 and LOADed that into the same table. then sampled my data.. surprise! after the first twenty entries (test1 and original test2) i expected to see test2's first entry in row 21 of my table but it was actually the sixth from last entry in test2! it filled in the last six entries as the first entries in the LOAD then continued to load the rest.. what gives?
t***@gmail.com
2009-06-10 20:36:07 UTC
Permalink
Also, i have done the LOAD with a bigger file...

i tried to load JANUARY data, so it should begin JAN 1, but the first row in the table after loading as primary key value of 315?? and the data is from JAN 5?? the seemingly semi-random loading of data seems to throw in huge chunks out of order, as this scenario happens elsewhere in the table. am i missing some configuration stuff or what??

sample LOAD command:

load client from "file-path" of del modified by identityignore messages "message-path" into TableName nonrecoverable
CRPence
2009-06-10 21:11:49 UTC
Permalink
FWiW: Physical ordering of data is moot to a relational database.
All that should matter for a data load operation, is that all of
the row data is there. Collation is something done in a run-time
SELECT according to the sort sequence of the environment and the
ORDER BY that was explicitly requested. Two tables with the same
keys and data, irrespective of the physical order of the rows, are
the same /set/ of data.

Regards, Chuck
Post by t***@gmail.com
Also, i have done the LOAD with a bigger file...
i tried to load JANUARY data, so it should begin JAN 1, but the
first row in the table after loading as primary key value of
315?? and the data is from JAN 5?? the seemingly semi-random
loading of data seems to throw in huge chunks out of order, as
this scenario happens elsewhere in the table. am i missing some
configuration stuff or what??
load client from "file-path" of del modified by identityignore
messages "message-path" into TableName nonrecoverable
Loading...