Find the number of times an SQL Index had been used

When you are building SQL DDL Indexes to improve performance it is also imperative to find Indexes that are not being used. Having identified those a decision can be made on whether to delete these unused indexes, or not.

Finding this information introduced me to a SQL View I had not used before: SYSTABLEINDEXSTAT

SYSTABLEINDEXSTAT contains the columns I want, the number of times the index has been used and the date it was last used. Alas, the view SYSINDEXES does not contain that information.

I always recommend that, on the partition you use, you run the following statement at least once to see all of the information that is available to you:

SELECT * FROM QSYS2.SYSTABLEINDEXSTAT LIMIT 10 ;

Using something better than DSPDBR

I am sure we have all used the Display Database Relations command, DSPDBR, to establish any file's or table's dependent objects. We can also use a SQL Table Function to give us more (better) columns of data than the DSPSBR command does.

But before I start giving examples of using this Table function I need something to work with. Let me start with a very simple DDL table:

01 CREATE TABLE MYLIB.TABLE1 02 (FIRST VARCHAR(20))

I do not need to insert data into the table.

Next I will create an Index over my Table:

Getting information about MTI using SQL

Maintained Temporary Indexes, MTI, are SQL Indexes that have been created automatically by the Db2 optimizer, without any outside intervention. The reason an MTI is created is there is no suitable existing Index to meet the requirements of a SQL query. MTI is just like any other SQL Index in my IBM i systems. MTI are temporary, as is suggested by their name, and are deleted when the partition is IPL-ed. A MTI might not be recreated the first time the SQL query is executed, it may take several times before the Db2 optimizer decides it is advantageous to create it. Therefore, to optimize your systems it is a good idea to review the Index Advisor on a regular basis, to determine if there are Indexes that should be created to alleviate the need for a MTI.

This begs the question: How can I see what MTI currently exist on my partition?

Fortunately there is a way using the MTI_INFO Table function.

The Table function has two optional parameters:

1. Table schema
2. Table name

A better way to find which file an Alias was built for

Earlier this year I described a way I could retrieve the schema/library and table/file that a SQL Alias had been built over. over the last few months I have discovered several other ways that will allow me to easily get to that information, and even the member too.

I tend to use SQL Alias for coping with SQL's inability to easily handle multi member files. I could use the Override Database command, OVRDBF, too but I prefer the ease of using the Alias as I can create, use, and then delete the alias all within the same program.

Before I start showing examples, I am going to need to create an Alias, ALIAS_3. Here I am going to create an Alias that will based on the source file DEVSRC, in my library MYLIB, and the source member TESTRPG:

CREATE OR REPLACE ALIAS MYLIB.ALIAS_3 FOR MYLIB.DEVSRC (TESTRPG)

Copying multiple spool files into one PDF

The task came in: I needed to take many spool files from an output queue and convert them into just one PDF. I have written about converting individual spool files into PDF, but now there was the need for the to be only one PDF.

My solution would need to:

1. Merge all of the spool files into one spool file
2. Convert the one spool file into a PDF in the IFS

Both of these I have done separately on many occasions; this was the first time to do them together.

I cannot use the spool files I used for this task, but I have two spool files in my personal output queue:

Which files are used by a Query?

This is a question I am asked enough for me to want to write this so I can direct the askers here. The question is:

How can I discover the files used by all the Query/400 objects in a library?

The Query/400 object is called a Query Definition, and is the following object type: *QRYDFN

To generate a list of the files they use is a two-step process.

The first process is to create an output file that contains the data. I use the Display Program Reference command, DSPPGMREF, to generate the output file. If I wanted to list all the files used by all of the Query definitions in the library MYLIB I would use the following:

Access System Directory information via SQL

I was asked for a list of all the enabled user profiles that are enrolled in the system directory on a partition. I know I could use the Display Directory Entry command, DSPDIRE, to create an outfile of the information:

DSPDIRE OUTPUT(*OUTFILE) OUTFILE(QTEMP/WOKFILE) DETAIL(*FULL) OUTFILFMT(*TYPE1)

Surely I would only want to retrieve the information for enabled profiles, or perhaps have a way to validate if a single user profile is enrolled in the system directory.

After some poking around I found a number of files in the library QUSRSYS that would give me the information I wanted. As the data is in files I can use SQL to get data and format the results in a way I would want, in real time.

Merlin comes to 7.3

When the announcement for IBM i 7.5 was made we were introduced to Merlin. This was made available for IBM i 7.4 TR6 too.

I do not remember there being any mention of Merlin being made available for earlier releases. It appears that I missed something as earlier this month IBM announced that Merlin would be available for IBM i 7.3 too.

For more information you will find the announcement here.

Personally I think this is a great move by IBM to broaden the number of their customers who can use Merlin, as there still are a lot of companies still running IBM i 7.3 on their partitions.

Find how to update the long comment column

Those of you who have used the SYSCOLUMNS and SYSTABLES views must have noticed that there is a column called LONG_COMMENT. On the partitions I work on this column is usually null. What is purpose of this column? And how is it updated?

A quick search in IBM's documentation introduced me to the COMMENT ON SQL statement. Before I give examples of how to use it, I am going to need a SQL DDL table to use:

01 CREATE OR REPLACE TABLE MYLIB.FIRST_TABLE 02 FOR SYSTEM NAME "TABLE1" 03 (FIRST_COLUMN FOR "FIRST" VARCHAR(20), 04 SECOND_COLUMN FOR "SECOND" VARCHAR(20)) 05 ON REPLACE DELETE ROWS ; 06 LABEL ON COLUMN FIRST_TABLE ( 07 FIRST_COLUMN IS 'First col', 08 SECOND_COLUMN IS 'Column heading' 09 ) ; 10 LABEL ON COLUMN FIRST_TABLE ( 11 FIRST_COLUMN TEXT IS 'First column', 12 SECOND_COLUMN TEXT IS 'Column text' 13 ) ; 14 LABEL ON TABLE FIRST_TABLE IS 'This is the first table' ;

New Power10 Servers announced

September last year IBM announced the first of the Power servers that included the Power10 chips, the E1080.

Today more Power10 based servers have been announced. All of these are smaller than the "flagship" E1080, but still pack considerable power (pun intended).

Click on image to see larger version

I am sure all of my regular readers know that I am "software" guy, not "hardware". Rather than embarrass myself, by not explaining the new servers features properly, I am giving you the links I have found about these new servers.

• IBM Newsroom: IBM Expands Power10 Server Family to Help Clients Respond Faster to Rapidly Changing Business Demands
• Redbook for Power10 servers
• Power10 documentation
• Individual server specifications
  • S1014 server
  • S1022 server
  • S1024 server

A big disappointment is that the newly announced E1050 will not run the IBM i operating system.

New SQL Table Function for Activation Group information

New release IBM i 7.5 and the corresponding Technology Refresh for 7.4, TR6, has given us a SQL Table Function that allows us to see information about the active activation groups within a job. In the past I have been able to get this information using the Work With Job command, WRKJOB:

WRKJOB JOB('*') OPTION(*ACTGRP)

The output options are limited to display ( * ) or printed output ( *PRINT ). For years I have taken printed output from commands and broken them apart into files either using SQL or RPG. It is nicer to have a View or Table Function where I do not have to do this anymore.

Change to Create Table statement to stop accidental deletion of the Table

It is always a danger that someone will accidentally drop (delete) the wrong file, not maliciously just accidentally.

01 CREATE TABLE MYLIB.TABLE1 02 (COLUMN1 CHAR(1)) ; 03 DROP TABLE MYLIB.TABLE1 ; Statement ran successfully

Within the new release of IBM i 7.5 is an addition to the Create Table statement that can prevent this. This does appear to be only in 7.5, I cannot find a mention of this in the documentation for IBM i 7.4 TR6.

This addition to the Create Table statement is: WITH RESTRICT ON DROP