From everything that I have found (and I am still totally confused by this),
table partitioning seems to be more about busting up the storage of a single
member into multiple members based on a field in the table as opposed to
having multiple standalone members in a PF. This seem more like a bunch of
LFs with select clauses.
From the doc at https://www.ibm.com/docs/en/ssw_ibm_i_73/pdf/rzaf3pdf.pdf,
From page 1
There are two methods for partitioning data:
1. Partitioned tables - The data for a table is spread across multiple
members within a single physical file
on a single IBM i instance. This is the recommended method for physical
partitioning of data in a table.
2. Distributed files (distributed tables) - The data for a table is spread
across multiple IBM i instances.
This option is no longer recommended for use.
And from page 22
Or, to partition PAYROLL by range, use the following code:
CREATE TABLE PRODLIB.PAYROLL
(EMPNUM INT,
FIRSTNAME CHAR(15),
LASTNAME CHAR(15),
SALARY INT)
PARTITION BY RANGE(EMPNUM)
(STARTING FROM (MINVALUE) ENDING AT (500) INCLUSIVE,
STARTING FROM (501) ENDING AT (1000) INCLUSIVE,
STARTING FROM (1001) ENDING AT (MAXVALUE))
This statement results in a table that contains three partitions. The first
partition contains all rows where
EMPNUM is less than or equal to 500. The second partition contains all rows
where EMPNUM is between
501 and 1000 inclusive. The third partition contains all rows where EMPNUM
is greater than or equal to
1001.
Someone please correct me and include a link to some good IBM i
documentation on this.
-----Original Message-----
From: MIDRANGE-L <midrange-l-bounces@xxxxxxxxxxxxxxxxxx> On Behalf Of Gad
Miron
Sent: Saturday, September 17, 2022 8:38 AM
To: midrange-l@xxxxxxxxxxxxxxxxxx
Subject: Table partitioning in 7.5
Hello sages
Having read that MultiSystem is a no-charge option in 7.5 I would very much
like to test table partitioning and hopefully use it.
Rational:
We heavily use Multi-Member-transaction-Files, Physical & Logical (several
hundreds files) Some such files have hundreds MBRs and a few over a
thousand.
These files are managed by ADDPFM / ADDLFM / OVRDBF and such.
This is nice and dandy for day to day operations but makes for a difficult
task querying and analyzing historical data.
So we regularly copy those files to a Single-MBR DataWarehouse-like files
adding the Member Name as a field (and a key) and use those DWH files for
historical analysis.
Several questions come to (my) mind when considering replacing it with table
partitioning:
1. How many partitions are allowed in a partitioned table or index?
2. After a partitioned table is created, can one use
ADDPFM/RMVM/CLRPFM/ADDLFM on it ?
(I've read that OVRDBF works with it) 3. Is there a performance penalty
that comes with partitioned tables ?
All answers will be thankfully received.
Gad
--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@xxxxxxxxxxxxxxxxxx To subscribe,
unsubscribe, or change list options,
visit:
https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxxxxxxxx
Before posting, please take a moment to review the archives at
https://archive.midrange.com/midrange-l.
Please contact support@xxxxxxxxxxxxxxxxxxxx for any subscription related
questions.
Help support midrange.com by shopping at amazon.com with our affiliate link:
https://amazon.midrange.com
midrange.com
