Book Image

Mastering KVM Virtualization

Book Image

Mastering KVM Virtualization

Overview of this book

A robust datacenter is essential for any organization – but you don’t want to waste resources. With KVM you can virtualize your datacenter, transforming a Linux operating system into a powerful hypervisor that allows you to manage multiple OS with minimal fuss. This book doesn’t just show you how to virtualize with KVM – it shows you how to do it well. Written to make you an expert on KVM, you’ll learn to manage the three essential pillars of scalability, performance and security – as well as some useful integrations with cloud services such as OpenStack. From the fundamentals of setting up a standalone KVM virtualization platform, and the best tools to harness it effectively, including virt-manager, and kimchi-project, everything you do is built around making KVM work for you in the real-world, helping you to interact and customize it as you need it. With further guidance on performance optimization for Microsoft Windows and RHEL virtual machines, as well as proven strategies for backup and disaster recovery, you’ll can be confident that your virtualized data center is working for your organization – not hampering it. Finally, the book will empower you to unlock the full potential of cloud through KVM. Migrating your physical machines to the cloud can be challenging, but once you’ve mastered KVM, it’s a little easie.
Table of Contents (22 chapters)
Mastering KVM Virtualization
Credits
About the Authors
About the Reviewers
www.PacktPub.com
Preface
Index

Tuning CPU and memory with NUMA


Before we start tuning CPU and memory for NUMA-capable systems, let's see what NUMA is and how it works.

What is NUMA?

NUMA is an abbreviation for Non Uniform Memory Access:

Figure 25: Reference from Wikipedia

Think about NUMA as a system where you have more than one system bus, each serving a small set of processors and associated memory. Each group of processors has its own memory and possibly its own I/O channels. It may not possible to stop or prevent access across these groups. Each of these groups is known as a NUMA node.

In this concept, if a process/thread is running on a NUMA node, the memory on the same node is normally called local memory and memory residing on another node is known as foreign/remote memory. This implementation is different from the SMP (Symmetric Multiprocessor System), where the access time for all of the memory is the same for all the CPUs.

There exists something called the NUMA ratio, a measure of how quickly a CPU can access local...