Book Image

Hyper-V 2016 Best Practices

By : Romain Serre, Benedict Berger
Book Image

Hyper-V 2016 Best Practices

By: Romain Serre, Benedict Berger

Overview of this book

Hyper-V Server and Windows Server 2016 with Hyper-V provide best-in-class virtualization capabilities. Hyper-V is a Windows-based, very cost-effective virtualization solution with easy-to-use and well-known administrative consoles. This book will assist you in designing, implementing, and managing highly effective and highly available Hyper-V infrastructures. With an example-oriented approach, this book covers all the different tips and suggestions to configure Hyper-V and provides readers with real-world proven solutions. This book begins by deploying single clusters of High Availability Hyper-V systems including the new Nano Server. This is followed by steps to configure the Hyper-V infrastructure components such as storage and network. It also touches on necessary processes such as backup and disaster recovery for optimal configuration. The book does not only show you what to do and how to plan the different scenarios, but it also provides in-depth configuration options. These scalable and automated configurations are then optimized via performance tuning and central management ensuring your applications are always the best they can be.
Table of Contents (15 chapters)
Hyper-V 2016 Best Practices
Credits
About the Authors
About the Reviewer
www.PacktPub.com
Preface

Performance tuning


After establishing performance counter baselines, it's time to interpret them. The values of networking, disks, and memory are self-explanatory, so let's go into the details of CPU sizing.

If the logical processor counter is low, but the virtual processor counter is high, it means that you can add more vCPUs to your virtual machines, as the logical processors are still available.

Theoretically, there is really no upper limit to how many virtual CPUs you can assign to virtual machines. The Microsoft recommendation is to not exceed more than 8 virtual CPUs per physical CPU core for server workloads and more than 12 virtual CPUs per physical CPU core for VDI workloads. However, there is no support limit, and there are low-workload scenarios where this recommendation can be extended.

My real-world experience from working with performance counters and baselines is to use a 1:4 ratio for production workload as a rule of thumb and a 1:12 ratio for Test/VDI workloads as a sizing...