Discussion:
Faster LOAD
(too old to reply)
Gregor Kovač
2017-09-27 09:14:26 UTC
Permalink
Hi!

I'm using EXPORT/LOAD to transfer a data for a table from one machine to another.
Table has almost 20 million rows, exported is 6 GB in size. Using LOAD takes about 3.5 hours. Is there a faster way to transfer a copy of a table from one machine to another one? At the end the target table must be identical to the source.

Best regards,
Gregor
j***@calculo-sa.es
2017-09-28 06:36:51 UTC
Permalink
Post by Gregor Kovač
Hi!
I'm using EXPORT/LOAD to transfer a data for a table from one machine to another.
Table has almost 20 million rows, exported is 6 GB in size. Using LOAD takes about 3.5 hours. Is there a faster way to transfer a copy of a table from one machine to another one? At the end the target table must be identical to the source.
Best regards,
Gregor
Have you considered to try Data Ingest? It first appeared on version 10.1
I'm enclosing two articles about it.

https://www.ibm.com/developerworks/data/library/techarticle/dm-1304ingestcmd/index.html

http://www.db2dean.com/Previous/DB2INGEST.html
Gregor Kovač
2017-09-28 11:30:38 UTC
Permalink
I've been testing INGEST today, but it is not faster.
It is a fascinating concept, but it is not faster than LOAD.
j***@calculo-sa.es
2017-09-29 07:17:38 UTC
Permalink
Post by Gregor Kovač
I've been testing INGEST today, but it is not faster.
It is a fascinating concept, but it is not faster than LOAD.
Good to know. According to the docs it said it was amazingly fast. Perhaps it was not so fast after all.
Gregor Kovač
2017-09-29 13:13:50 UTC
Permalink
Docs on developerWorks states that INGEST is fast, but LOAD is very fast. You can find this in section "Comparing ingest to other methods".
Luiz da Silva
2017-10-09 13:44:35 UTC
Permalink
Post by Gregor Kovač
Docs on developerWorks states that INGEST is fast, but LOAD is very fast. You can find this in section "Comparing ingest to other methods".
At your destination database, you can try to load from a remote (cataloged) database without having the intermediate steps of exporting to flat, moving to destination and then loading.

Example:

declare C1 cursor database <SOURCE_DATABASE> user <USERNAME> using <PASSWORD> for select * from <SOURCE_SCHEMA>.<TARGET_TABNAME> ;

load from C1 of cursor replace into <TARGET_SCHEMA>.<TARGET_TABNAME> nonrecoverable ;
Loading...