Solved

CPU usage

Posted on 2013-06-12
4
701 Views
Last Modified: 2013-06-13
Hi -

We have a IBM AIX server using LPAR. This has 9 physical cpus (entitled) and 80 logical cpus. The logical CPU % of Utilization is never matching with the physical cpu utilization. When Physical CPU maxed out the logical CPU still showing only 30% of usage.

Please find the attached excel sheet for more details.

IF you look at the LPAR tab in the given excel, when the entitled cpu usage maxed out, the logical and physical should report the same % of CPU utilization.

Thanks!

Sve.
Book15.zip
0
Comment
Question by:sventhan
[X]
Welcome to Experts Exchange

Add your voice to the tech community where 5M+ people just like you are talking about what matters.

  • Help others & share knowledge
  • Earn cash & points
  • Learn & ask questions
  • 2
  • 2
4 Comments
 
LVL 62

Expert Comment

by:gheist
ID: 39242776
Aix server model please...
uname -r
And explain what problem you are trying to solve.

 physical CPU is hared between virtual CPUs, and virtual one is waiting for IO while rest of
physical CPU power is used by other virtual CPUs. the numbers will never add up

Please unzip document yourself, i am not good at peeling onions.
0
 
LVL 18

Author Comment

by:sventhan
ID: 39242927
Thanks for your time. I've peeled the onion for you.

The Question is,

How 1 Physical CPU match with the X logical CPU?  How this is getting configured?

When all the CPUs being used , PCPU = 100% why lcpu is only 30%?

What is the ratio between these 2 cpus?

Thanks,
sve.
Book15.xlsx
0
 
LVL 62

Accepted Solution

by:
gheist earned 500 total points
ID: 39243096
100/30 - ratio is 333,3%

Pardon - but what PROBLEM you are trying to solve?

100% PCPU = 100% LCPU
LCPU usage can be less because:
VCPU is guaranteed num_pcpus/num_vcpus of cpu time (in your case - 9/80=0.113=11,3% of a single CPU)
If one does nothing (or waits for slow network IO) other VCPUs can use spare time
If enough of them do nothing system can save power by halting and slowing down unused PCPUs.

Remember that not all workloads are calculations on CPU.
Database would wait a lot for disk IO
While webserver will wait for network io
Or websphere will spend huge amounts of time blocking and synchronizing threads
All three showing as "wait" and/or "idle" in topas
0
 
LVL 18

Author Closing Comment

by:sventhan
ID: 39245641
Thanks
0

Featured Post

What Is Transaction Monitoring and who needs it?

Synthetic Transaction Monitoring that you need for the day to day, which ensures your business website keeps running optimally, and that there is no downtime to impact your customer experience.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

Using libpcap/Jpcap to capture and send packets on Solaris version (10/11) Library used: 1.      Libpcap (http://www.tcpdump.org) Version 1.2 2.      Jpcap(http://netresearch.ics.uci.edu/kfujii/Jpcap/doc/index.html) Version 0.6 Prerequisite: 1.      GCC …
Join Greg Farro and Ethan Banks from Packet Pushers (http://packetpushers.net/podcast/podcasts/pq-show-93-smart-network-monitoring-paessler-sponsored/) and Greg Ross from Paessler (https://www.paessler.com/prtg) for a discussion about smart network …
Learn how to find files with the shell using the find and locate commands. Use locate to find a needle in a haystack.: With locate, check if the file still exists.: Use find to get the actual location of the file.:
How to Install VMware Tools in Red Hat Enterprise Linux 6.4 (RHEL 6.4) Step-by-Step Tutorial

717 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question