Sizing

Edit online

Accurate workload sizing is a critical prerequisite for running Red Hat® OpenShift® workloads on IBM Z® and IBM® LinuxONE. This document highlights why systematic sizing is essential to achieve optimal performance for containerized workloads and explains how to do sizing for Red Hat OpenShift deployments on IBM Z and IBM LinuxONE.

Workload sizing is the task of determining the right set of resources to run a specific workload reliably and efficiently. Many factors influence this process, including the configuration of Red Hat OpenShift Container Platform, the Red Hat OpenShift add-ons in use, and the workload requirements. Required resources include compute, memory, and storage. The main factors are highlighted and you are shown how to use them to achieve a reliable overall sizing for the environment.

This document is specific for Red Hat OpenShift Container Platform (RHOCP) running on IBM Z and IBM LinuxONE. RHOCP is Red Hat's enterprise Kubernetes distribution. It provides developers, IT organizations, and business leaders with a hybrid cloud application platform for developing and deploying new containerized applications on a secure platform.

Introduction

The IBM Z family comes with a powerful computation resource, the Integrated Facility for Linux (IFL). On IBM Z, IFL cores are dedicated to running the Linux operating system. IFLs can be used with different hypervisor options. For example, z/VM® or KVM.

Red Hat provides an extensive set of sizing documents that mainly focus on non-IBM Z environments. Mapping the virtual CPU (vCPU) requirements to IFLs is unfeasible, since it depends on the required hypervisor and architecture.

The objective of this document is to give an estimation of the required IFL cores for a given workload. By focusing on IFL requirements, it can be considered as an addition to the Red Hat sizing guides for vCPU and memory.

The overall load on a cluster as the sum of static and dynamic load is defined as follows:

O C P C l u s t e r_{I F L} = S t a t i c_{I F L} + D y n a m i c_{I F L}

$S t a t i c_{I F L}$: The static load refers to the operator demand of a Red Hat OpenShift cluster for a given operator.
$D y n a m i c_{I F L}$: The dynamic load refers to the additional demand of compute resources for a given design.