21.02.2013 Views

AIX 5L Problem Determination - IBM Redbooks

AIX 5L Problem Determination - IBM Redbooks

AIX 5L Problem Determination - IBM Redbooks

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

later). The method used in <strong>AIX</strong> Version 4.3.2 and earlier versions of the<br />

operating system can, under certain circumstances, give an inflated view of wa<br />

time on SMPs. In <strong>AIX</strong> Version 4.3.2 and earlier, at each clock interrupt on each<br />

processor (100 times a second per processor), a determination is made as to in<br />

which of the four categories (usr/sys/wio/idle) to place the last 10 ms of time. If<br />

any disk I/O is in progress, the wa category is incremented. For example,<br />

systems with just one thread doing I/O could report over 90 percent wa time<br />

regardless of the number of CPUs it has.<br />

The change in <strong>AIX</strong> Version 4.3.3 is to only mark an idle CPU as wa if an<br />

outstanding I/O was started on that CPU. This method can report much lower wa<br />

times when just a few threads are doing I/O and the system is otherwise idle. For<br />

example, a system with four CPUs and one thread doing I/O will report a<br />

maximum of 25 percent wa time. A system with 12 CPUs and one thread doing<br />

I/O will report a maximum of 8.3 percent wa time.<br />

Also, NFS now goes through the buffer cache, and waits in those routines are<br />

accounted for in the wa statistics.<br />

A wa value over 25 percent could indicate that the disk subsystem might not be<br />

balanced properly, or it might be the result of a disk-intensive workload.<br />

The faults columns<br />

It may also be worthwhile to look at the faults columns, which provide information<br />

about process control, such as trap and interrupt rate.<br />

The in column<br />

In the in column is the number of device interrupts per second observed in the<br />

interval.<br />

The sy column<br />

In the sy column is the number of system calls per second observed in the<br />

interval. Resources are available to user processes through well-defined system<br />

calls. These calls instruct the kernel to perform operations for the calling process<br />

and exchange data between the kernel and the process. Because workloads and<br />

applications vary widely, and different calls perform different functions, it is<br />

impossible to define how many system calls per-second are too many. But<br />

typically, when the sy column raises over 10000 calls per second on a<br />

uniprocessor, further investigation is called for (on a SMP system, the number is<br />

10000 calls per second per processor). One reason for this high number of calls<br />

per second could be polling subroutines like select(). For this column, it is<br />

advisable to have a baseline measurement that gives a count for a normal sy<br />

value.<br />

240 <strong>IBM</strong> ^ Certification Study Guide - <strong>AIX</strong> <strong>5L</strong> <strong>Problem</strong> <strong>Determination</strong> Tools and Techniques

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!