database - DB2 load partitioned data in parallel -
I have 10-node DB2 9.5 database, with raw data on each machine (i.e. node 1: / scratch / data) / Dataset.1 node2: /scratch/data/dataset.2 ... node10: /scratch/data/dataset.10
There is no shared NFS mount - none of my machines have all the datasets
Each line in a dataset file is a long string text, column is delimited. The first column is the key. I do not know that the hash function which is subdit of DB2 Yoga will be done, so the dataset pre-
There is no shortage of renaming all my files, is there any way to load DB2 in parallel? I am trying to do this < / P>
'Dell scratches / data / dataset modified by cold; FastShare message / dev / null has been changed to TESTDB.data_table part_file_location' / scratch / data ';
It does not know how to suggest DB2 that it is on the first node If the Indivi Dual Data files are split from the same database partition on each partition. Was not produced, then you are stuck, and have to load 10 times - once each different database partition You can do this with db2_all to load it in the same command:
connect to db2_all "db2 db; / scratch/data/dataset.\$DB2NODE del from db2 load. .. "
Do not try to run db2_all command in parallel; -)
Another idea for the future: Do you have enough room on the same server if you include all files first? You can load with a nominated pipe:
mkfifo f cat dataset *. Gz | Gzip -dc & gt; F & amp; Load DB2 from "Dell to F ...."
Comments
Post a Comment