Uploaded image for project: 'OpenShift Request For Enhancement'
  1. OpenShift Request For Enhancement
  2. RFE-2030

Workload specific performance tunables and settings

    XMLWordPrintable

Details

    • 0
    • 0% 0%

    Description

      Certain workloads have a need to have custom performance tuning profiles applied.

      For e.g. the IBM Cloud Pak for Data requires various kernel parameters to be set:
      (vm.max_map_count) (kernel.msgmax, kernel.msgmnb, kernel.msgmni) (kernel.shmmax, kernel.shmall, and kernel.shmmni) (kernel.sem)

      More details can be found here: https://www.ibm.com/support/producthub/icpdata/docs/content/SSQNUZ_latest/cpd/install/node-settings.html
      And: https://www.ibm.com/docs/en/cloud-paks/cp-data/3.5.0?topic=tasks-changing-required-node-settings#node-settings__kernel

      The settings could be different depending on the underlying architecture, what the default page size is for that architecture, how many threads are available per core etc. For example for IBM Power, the default page size is 64k and the SMT settings could be varied between 1,2,4 and 8. This has an affect on the kernel tunables and other settings.

      Currently users have to install OpenShift and post-installation go through a laborious and manual process to prepare the system, before proceeding to Install the workload (Cloud Pak for Data). A custom performance profile per workload, that will allow for variability based on the underlying architecture would ease the burden for the administrators.

      Attachments

        Activity

          People

            rhn-support-dhardie Duncan Hardie
            manoj5 Manoj Kumar
            Votes:
            0 Vote for this issue
            Watchers:
            5 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: