Users from iSeries and System i often notice that their disk usage is skewed. Some disks are much busier than others. This extra traffic may arise from the fact that certain journal receivers are making deposits onto some disk arms more aggressively than others. Let us explore how journaling can contribute to this and what actions can be taken to minimize this skew.
Written by Larry Youngren and Christopher Kundinger
The facts of journal disk usage
The number of disk arms that journal receiver data is spread across is influenced by the journal receiver threshold value (THRESHOLD parameter on the CRTJRN command) along with the choice selected for the RCVSIZOPT parameter on the CHGJRN command. Any journal with a RCVSIZOPT setting of *MAXOPT1 or greater will examine the THRESHOLD setting before deciding how many disk arms to employ. The choices range from 10 disks up to 100.
The following equation is used when determining how many disk arms will be used:
Number of disk arms = journal receiver threshold setting / 64 MB
Why are some disks noticeably busier then others?
As discussed, the journal receiver threshold dictates the quantity of disk arms used. For example, if you had a journal receiver threshold of 600000, 10 disk arms would be used because 600000/64MB = 9.2, which is less than the minimum of 10. In an intense journal environment it would correspondingly be normal to see 10 disks more active than others. This would remain the case until a change journal was issued, allowing the new journal receiver to choose 10 different disk arms.
How many disk arms is my journal using?
It can be instructive to analyze your own journal receivers. By doing so, you can ascertain precisely how many disk arms the receivers are spreading themselves across.
The following instructions will tell you how many disks arms the journal you are analyzing is actually using. To ensure an accurate measurement, a significant number of journal entries should be gathered.
1) Display the journal to an outfile with a CL command similar to that shown below. It is important that you use the *TYPE5 option on the DSPJRN CL command.
DSPJRN JRN(LIB/JRN) OUTPUT(*OUTFILE) OUTFILFMT(*TYPE5) OUTFILE(LIB/OUTFILE)
2) Query the resulting outfile to determine the number of disk arms being used:
SELECT MAX(JOARM) FROM LIB/OUTFILE
Why am I not using as many disk arms as I expected?
Have some disks been recently added to your disk pool? Note that a journal receiver will not start using these recently added disks until the next CHGJRN command has been used to attach a new journal receiver.
How can I increase the number of arms being used by the journal?
Provided that you have specified a RCVSIZOPT of *MAXOPT1 or higher, increasing the threshold size for your journal receiver will give your journal permission to use more disk arms. This can be done via the THRESHOLD parameter on the Create Journal Receiver (CRTJRNRCV) command. If Version 5 Release 4 or later of the i5/OS™ operating system is installed, then increasing your threshold can also be done via the THRESHOLD parameter on the Change Journal (CHGJRN) command. By raising your threshold you will tend to reduce disk arm skew by causing your journal receiver to spread itself across more disk arms. In addition, the more disk arms offered to the journal, the wider the effective bandwidth at its disposal, which can be helpful when peaks of journal disk traffic ensue.
For more details about journaling and disk arm usage see the Information Center at:
This material has not been submitted to any formal IBM test and is published AS IS. It has not been the subject of rigorous review. IBM assumes no responsibility for its accuracy or completeness. The use of this information or the implementation of any of these techniques is a client responsibility and depends upon the client's ability to evaluate and integrate them into the client's operational environment. publib-b.boulder.ibm.com