In general no – there are no specific requirements. As long as the processor has more than one core, core parking will opportunistically park cores. Core parking is simply a mechanism to express to the scheduler a preference on which processors to use.  By virtue of the scheduler adhering to that preference, we are able to consolidate work on to fewer processors.  The “parked” processors which should have less work to do will likely transition to “Idle” sooner and for longer, thereby providing greater power savings.  Hence, there are no hardware requirements for the actual “parking/unparking” selection process.

“Parked” is not the same as “Idle”

Parking is a hint to the scheduler; parking promotes the chances of the core to enter idle. It does not imply idle; nor is it a pre-requisite for idle.

Power saved during “Idle” is highly dependent on HW/processor characteristics.

For example features such as C1E, C6 and others all potentially increase power savings- some processors have these some don’t.  Again, core parking does not directly influence how much power is saved during idle. It only promotes the chances of being idle.

The Core Parking engine does not park sockets/packages

All parking is done on a logical processor basis. If all cores that constitute a package get parked during the operation of the engine; it is coincidental. In fact, this is rare because the engine always leaves at least one logical processor per NUMA node unparked and so if a package/socket is a NUMA node, there will never be a case in which all cores in the socket are parked. Hence the only way all logical processors in a package get parked is if there are multiple sockets per NUMA node.

Resource Monitor shows the current parking status of each core.

There isn’t a way to know which cores/sockets are idle at a particular time instant. This information may not be very interesting because a processor is typically transitioning in and out of idle 100s of times a second. %idle time is in many cases the metric of interest.

In performance Monitor – Processor Information -> %Idle Time. You can select the logical processors of interest and the counter says what fraction of time the processor has been Idle. This performance counter gives information about the % of time each logical processor in the system has been idle. Unfortunately, there is no similar counter for sockets.

NetApp Storage Best Practices for Microsoft Virtualization, also known as TR-3702, has been updated for the release of Hyper-V R2 and SCVMM 2008 R2.  NetApp has added new sections covering the following:

• Hyper-V Server Networking Considerations
• Hyper-V Cluster Networking Considerations
• Hyper-V Server Storage Provisioning
• Virtual Machine Storage Provisioning
• NetApp FlexClones and Snapshot technology
• Backup and Recovery
• Disaster Recovery and High Availability

This guide now contains not just best practices, but additional content provided to educate the reader on how and why we established some of our best practices. I have received lots of great feedback on this document so far, mostly because it’s packed with everything that folks have been asking for to help them with their Hyper-V deployments.

But wait there’s more the following guides have been updated too:

• NetApp Implementation Guide for Microsoft Virtualization (TR-3733)

Find out who these customers are and how they are getting the benefits of Virtualisation while they save money.

Hit this link for these case studies https://www.microsoft.com/casestudies/Case_Study_Search_Results.aspx?Type=1&Keywords=virtualization&LangID=46

If you are a customer in Australia using Microsoft’s Virtualisation solutions and want to share your story with the rest of the world please email me apawar@microsoft.com

https://www.microsoft.com/downloads/details.aspx?FamilyID=64a5cc28-f8a1-4b30-a4a2-455c65bda8d7&displaylang=en

So here is a link for the How to Videos that the Virtualisation team has released https://www.microsoft.com/virtualization/en/us/demo.aspx

• Plan
• Install
• Best Practices
• Deployment
• Pre-Deployment Tools
• Management
• Performance
• Workload Specific Guidance

Microsoft now has one page that lists these resources https://technet.microsoft.com/en-us/dd565807.aspx

The guide addresses the following fundamental decisions and tasks:

• Identifying which applications and which desktops will be delivered by Remote Desktop Services.
• Determining whether to deliver desktops by using session-based presentation virtualization or Virtual Desktop Infrastructure (VDI).
• Designing the infrastructure needed to use Remote Desktop Services to serve the selected applications and desktops.

Download it from Windows Server 2008 R2 Remote Desktop Services

https://technet.microsoft.com/en-us/library/ee441285.aspx

Server Core

https://technet.microsoft.com/en-us/library/cc753802(WS.10).aspx

https://h20195.www2.hp.com/V2/GetPDF.aspx/4AA0-2391ENW.pdf

Australia is a strong adopter of virtualisation. In this market, there is a lot of information about Microsoft virtualisation technologies and it can be hard to know what’s true and what’s not. In this session we will discuss what each of the virtualisation vendors are delivering.

You will learn about:

• The future vision of virtualisation, not just in the next generation of products, but also in what direction virtualisation is heading.
• The evolution of integration components for this multi-vendor environment.
• The functionality to manager a multi-vendor environment that many customers are adopting.

Sydney
Date: 3 December, 2009
Time: 9:30am registration, session: 10:00-11:30am
Location: Microsoft, 1 Epping Road, North Ryde NSW
Click here to register

Melbourne
Date: 4 December, 2009
Time: 9:30am registration, session: 10:00-11:30am
Location: Microsoft, Level 5, 4 Freshwater Place, Southbank VIC
Click here to register