I was reading this new paper... There was a discussion on this list about vm.swappiness a while ago. We had found with it at 60, our biggest app would just continue to march through its swap space and keep increasing its size. We found things happier at 60. But I never did get a good feeling about what this setting really did.
This study recommends setting it to zero. <from paper> Setting swappiness to zero These instructions are used to set the swappiness parameter to zero. The swappiness parameter influences the kernel preference to move memory pages from applications to swap page, versus reclaiming memory from the cache. After system restart, set the swappiness parameter to zero. This ensures that if memory is constrained, the page cache is reduced in an attempt to recover memory, before application pages are moved to swap space: +--------------------------------------------------------------+ |echo 0 >/proc/sys/vm/swappiness +-------------------------------------------------------------+ This setting may improve or degrade the performance of an application. Since there is adequate memory already dedicated to this workload, large amounts of memory would not need to be swapped to disk anyway. Since precautionary 'early' swapping is now avoided, our results should be free of the effects of this kind of swapping. </from paper? Wouldn't this always be a good thing under VM to force it to reduce its memory footprint? Are there cases where it wouldn't be? Inquiring minds... Marcy "This message may contain confidential and/or privileged information. If you are not the addressee or authorized to receive this for the addressee, you must not use, copy, disclose, or take any action based on this message or any information herein. If you have received this message in error, please advise the sender immediately by reply e-mail and delete this message. Thank you for your cooperation." -----Original Message----- From: Linux on 390 Port [mailto:[email protected]] On Behalf Of Dorothea Matthaeus Sent: Monday, September 21, 2009 4:40 AM To: [email protected] Subject: [LINUX-390] New Performance Whitepaper:: WebSphere on IBM System z 64-bit/31-bit Studies with J2EE Workloads WebSphere on IBM System z 64-bit/31-bit Studies with J2EE Workloads The paper is available from : DeveloperWorks http://www.ibm.com/developerworks/linux/linux390/perf/tuning_pap_websphere.html#j2ee This study explores the performance of a WebSphere Application Server 6.1 system under a customer-like J2EE application workload. It includes a description of how the test environment was set up and how the systems were configured. The difference in performance behavior of the 31-bit and the 64-bit WebSphere versions are compared, and the impact of heap size and garbage collection are analyzed. Dorothea Matthaeus Linux on System z Information Development IBM Deutschland Entwicklung GmbH ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
