Re: [ITM62] Internal CPU Utilization Situations
John Alvord <jalvord-r/[email protected]>
| Newsgroups | gmane.comp.sysutils.tivoli.general |
|---|---|
| Message-ID | <OFECF6E148.5404B605-ON88257B94.0064063F-88257B94.0065A479__36731.0323672884$1372098740$gmane$org@us.ibm.com> |
Not sure what the SLA is intended for... but counting I/O wait time is an interesting idea. If you imagined an extreme case where user/system was 5% and I/O Wait time as 95% - you are really at the end of your resources... any new work added [of the same type] would not be able to progress. In that case you might need better disk I/O or maybe more memory... but it clearly needs attention just like a lack of CPU might need more CPU resources. Regards, ______________________________________________ John Alvord Advisory Engineer IBM Tivoli Monitoring 6 Platform Customer Support Office: 1-720-396-2788 Mobile: none jalvord-r/[email protected] Cloud & Smarter Infrastructure IBM Support Portal Open a New Service Request Update your PMR via Email Escalate your PMR My Blog: Wonderful World of Situations Secure upload to ECUREP From: "[email protected]" <[email protected]> To: saby <sabyantony2001-/[email protected]>, "Discussion list for Tivoli product and Tivoli Ready products." <[email protected]>, Date: 06/24/2013 10:52 AM Subject: Re: [TME10] [ITM62] Internal CPU Utilization Situations Sent by: [email protected] Also, a quick peek in the 6.3-level UX agent User's Guide shows that "CPUBUSY" is "User CPU (Percent) + "System CPU (Percent)". Looking in turn at each of those(USERCPU/SYSCPU), it shows they are both derived from the mpstat command output for the 'us' and 'sy' columns. So it omits the IO Wait CPU usage then (mpstat command's 'wt' column). That should give you enough to know whether it will suffice for purposes of proving out SLAs for your customers. Toben On Mon, Jun 24, 2013 at 11:19 AM, saby <sabyantony2001-/[email protected]> wrote: From: saby <sabyantony2001-/[email protected]> To: "[email protected]" <[email protected]>; Discussion list for Tivoli product and Tivoli Ready products. <[email protected]> Sent: Monday, June 24, 2013 9:39 PM Subject: Re: [TME10] [ITM62] Internal CPU Utilization Situations Thanks for the info. I also used the same formula, (100- Idle%), to get the CPU usage. But this output is different than optium report. Now my manager is asking me to explain the difference between ITM and optium report. I'm using SMP CPU attributes . Is there any issue with this? (AIX system has 20 cores.) Regards Sebastian From: "[email protected]" <[email protected]> To: saby <sabyantony2001-/[email protected]>; Discussion list for Tivoli product and Tivoli Ready products. <[email protected]> Sent: Monday, June 24, 2013 9:27 PM Subject: Re: [TME10] [ITM62] Internal CPU Utilization Situations I agree w/Jared about the superior metrics from the PX agent, especially as it regards true core usage--((sys % + user %)/100) * Num of Physical Cores--but from the sar report comparison I did with out AIX admins UX's CPU Busy % attribute tracked well over time. I think the CPU Busy % is akin to (100 - Idle %), or put another way, very close to (User %+ System %+ IO Wait%) from what I recall. On Mon, Jun 24, 2013 at 10:36 AM, saby <sabyantony2001-/[email protected]> wrote: Hi List, I want to elaborate the scenario. As per SLA if CPU utilization is more than 70% for a sustained period of 15 mintues then it is a breach. So which CPU attribute will help me to acheive this? Can anyone guide me on this? Agent OS -- AIX ITM -- 6.2.3 FP3 Report -- custom cognos report Regards Sebastian From: saby <sabyantony2001-/[email protected]> To: Discussion list for Tivoli product and Tivoli Ready products. < [email protected]> Sent: Monday, June 24, 2013 6:28 PM Subject: Re: [TME10] [ITM62] Internal CPU Utilization Situations Thanks Jared. I'm using 6.23 FP3 and have to generate CPU reports for SLA (>70%). CPU Busy (percent) is breaching most of the times and we did a comparison with optim . Optim shows 40% when CPU Busy is 70% in ITM. So I want to know for SLA reports which attribute is best? What is the purpose of following attributes? Avg CPU Busy 5 CPU Usage Regards Sebastian From: Jared Seaton <[email protected]> To: saby <sabyantony2001-/[email protected]>; Discussion list for Tivoli product and Tivoli Ready products. <[email protected]> Sent: Monday, June 24, 2013 6:13 PM Subject: Re: [TME10] [ITM62] Internal CPU Utilization Situations I find that on AIX, it is best to use either the AIX Premium agent, or Unix OS Agent 6.2.3 FP2+ and monitoring Entitlement metrics, over standard Unix CPU metrics. We have situations that monitor the Entitlement Used Pct, and Phys Busy Pct attributes of the AIX Premium agent. The AIX metrics were introduced into the Unix OS Agent at 6.2.3 FP1, but I found too many issues at that level, and wouldn't recommend it. ------ Jared Seaton twitter.com/jmseaton On Mon, Jun 24, 2013 at 8:15 AM, saby <sabyantony2001-/[email protected]> wrote: Hi, I'm facing almost same issues in CPU utilization (AIX ). Which is the correct attrribute to analyze CPU utilization? Polling interval is 5 mints and I used CPU Busy (percent) attribute. But when I compare with optim / topaz reports, ITM is showing higher values (almost double). There are multiple attributes available with the agent. Which one to use , I'm confused . Can anyone help me? Avg CPU Busy 1 Avg CPU Busy 5 Avg CPU Busy 15 CPU Busy (Percent) CPU Usage Regards Sebastian From: "Hetter, Larry" <[email protected]> To: "[email protected]" <[email protected]> Sent: Friday, September 25, 2009 12:43 AM Subject: [TME10] [ITM62] Internal CPU Utilization Situations I am running ITM 6.2 FP2 using some of the built-in CPU situations for the SMP CPU attribute group on AIX boxes. When I go to the System Information and System Details screen in TEPS, CPU ultization does not always seem to reflect the same things that the sar command and topas command show when they are run on the same box. The internal CPU Situation using the Average CPU Busy Attribute Item also occasionally fires when other indicators on the box do not reflect high CPU utilization. Has anyone else noticed this? Does anyone know the behind the scenes commands used by TEPS to determine CPU utilization? Thanks. Larry Hetter CareFirst BlueCross BlueShield UNIX Systems and Event Management 410-880-6543 [email protected] ******************************************************************************** Unauthorized interception of this communication could be a violation of Federal and State Law. This communication and any files transmitted with it are confidential and may contain protected health information. This communication is solely for the use of the person or entity to whom it was addressed. If you are not the intended recipient, any use, distribution, printing or acting in reliance on the contents of this message is strictly prohibited. If you have received this message in error, please notify the sender and destroy any and all copies. Thank you. ******************************************************************************* _______________________________________________ TME10 mailing list [email protected] Unsubscribe:[email protected] _______________________________________________ TME10 mailing list [email protected] mailto:Unsubscribe%[email protected] _______________________________________________ TME10 mailing list [email protected] Unsubscribe:[email protected] _______________________________________________ TME10 mailing list [email protected] mailto:Unsubscribe%[email protected] -- "Sometimes I think that's the only right thing to do: To dream. to live in the world of dreams. But it doesn't last forever--wakefulness always comes to take me back..." _______________________________________________ TME10 mailing list [email protected] Unsubscribe:[email protected] -- "Sometimes I think that's the only right thing to do: To dream. to live in the world of dreams. But it doesn't last forever--wakefulness always comes to take me back..." _______________________________________________ TME10 mailing list [email protected] Unsubscribe:[email protected] _______________________________________________ TME10 mailing list [email protected] Unsubscribe:[email protected]