Maximizing Virtual Machine Performance - resource paper to get the most of your infrastructure by doing some performance tuning.
In this technical Whitepaper called Maximizing Virtual Machine Performance, Mattias Sundling walks you in several steps, through the basics on where you can recover the maximum of performance of your VMs, which is otherwise lost in missconfigurations.
A well-tuned foundation enables you to make better use of your virtual infrastructure. Optimizing CPU, memory, disk and network will improve performance and make your virtual environment more efficient to manage. Download this white paper by Quest Evangelist, Mattias Sundling.
Maximizing Virtual Machine Performance – The Outline:
- Hardware Version 8 advantages over Hardware v. 7.
- vCPU and HAL (Hardware Abstraction Layer) config caveats and missconfigurations.
- CPUReady, what’s the metric for?
- Virtual NUMA
- Memory Limits, the sizing process and it’s caveats. The definitions (granted, Ballooned, Active, swapped…)
- Memory reclamation (TPS)
- SSD disk swap cache
- LUN sizing (what happens when your LUNs are too big – SCSI reservation conflicts, lower I/O due a metadata locking
- VMFS and guest OS alignement, SIOC (storage I/O control)
- Network I/O control, vMotion, sVmotion, what’s the improvement in vSphere 5.
A quick quote:
Make sure you select the correct hardware abstraction layer (HAL) in the guest OS. The HAL drives the OS for the CPU; choices are “Uni-Processor (UP) single processor” or ”Symmetric Multiprocessing (SMP) multiple processors”.
If you have a VM running Windows 2003 SP2 or later that has been reduced from two vCPU to one vCPU, you will still have the multiprocessor HAL in the OS. This results in slower performance than a system with correct HAL. The HAL driver can be manually updated; however, Windows versions prior to Windows 2003 SP2 cannot be easily corrected. I have personally experienced systems with an incorrect HAL driver. They consume more CPU, which can often peak to unnecessarily high CPU-utilization percentages when the system gets stressed.
Mattias is vExpert and he is known for his deep technical expertise in server virtualization, which he regularly shares in such different formats as blog posts and white papers, and as a speaker at industry events including VMworld U.S. and Europe, Virtual Forum, and VMUGs across the U.S. and Europe.
I can only highly recommend to read this paper from Mattias Sundling. Download here: Maximizing Virtual Machine Performance
Mattias also presented the very interesting VMworld Session together with Eric Sloof. The Session that I blogged about recently is called : Mythbusters Goes Virtual. You do have a possibility to download the session whitepaper as well.
Maximizing Virtual Machine Performance was published on ESX Virtualization.