VIR308Hyper-V and Dynamic Memory in DepthName: Bryon Surace Title: Senior Program Manager

Name: Senthil Rajaram Title: Senior Program Manager

Microsoft Corporation

Agenda

Memory & Virtualization Memory Optimization TechniquesHyper-V Dynamic Memory

Architecture & ConceptsDemoSystem Impact

Q&A

Memory & Virtualization

Virtualization & MemoryHow much memory does a server actually need?

IIS Server?Print server?File server?Branch Cache?Direct Access?

How much will performance be affected if you halved the amount of memory in a VM?

No one wants to size VMs Memory

“New virtual machines get 1GB of RAM [no matter what the VM is running]. I only give people more memory if they complain about performance”

“All VMs get 4GB of RAM [I have no idea what is happening with that memory] and no one complains”

“I take the minimum system requirements and add (insert one: 50%, 100%, 150%)”

“A vendor tells me their app needs 4GB of RAM. I do not have the time to test this to find out if it is true or not”

Virtualization & MemoryMemory

Key Factor to the number of running VMsThe most utilized asset in the system, but also a more expensive asset

Statistics on resource utilization of workloads CPU 10%Memory 40%Network I/O <5%Disk I/O <5%

Customer RequirementsMaximum density, without sacrificing performanceMaintain consistent performanceDon’t provide a feature that’s unsuitable for production use

Dynamic Memory GoalsHigher VM consolidation ratios with minimal performance impact

Dependent on:How much variation in memory utilization the workloads haveHow good a job you did of sizing the systems in the first place

Work well for both server and desktop workloadsAdd minimal overhead to the systemPass the “that looks right” test

Understanding philosophical differences

Host and Guest CollaborateUnderstands what guest information to useTrying to get the “best bang for buck” in virtual memory management

Host works in IsolationDoes not trust guest information

Building a “black box” solution

Dynamic Memory, not Overcommit

Overloaded TermOver-subscription vs. Over-committment

No one wants to overcommit their resourceYou don’t overcommit other resources (really – you do not)VMware does not want you to overcommit memory (really)

Dynamic Memory treat memory like we treat CPU resources

Dynamically schedulable resource

Memory Management & Optimization Techniques

How it works:Increasing the size of the balloon forces the guest to react to memory pressure by releasing unused pagesDecreasing the size makes more memory available to the guest

Ballooning

VM

Memory

Host Computer

Balloon

Disk

VM

Host Computer

Memory

Balloon

DiskInflate

Swap Out

Ballloon

Deflate

VM

Host Computer

Memory

Disk

Swap In

Balloon

Deflate

Second Level PagingHow it works:

Paging at the virtualization layer by swapping VMs memory out to disk

Many problems:Swapping Guest Kernel ResourcesDouble PagingDisks are slow

But it always works…

External Page SharingHow it works:

Eliminate redundant copies of memory pages common to more than one virtual machine1. Hash all memory and store it in a table…2. Identify the common hashes and then…3. Perform a bit by bit comparison

ProblemsPage Sharing not dynamicCan take hours to share pagesThe largest benefit are zero pagesDoesn’t work with large pages

Other TechniquesGuest directed page sharingMemory compressionAnd on…

Hyper-V R2 SP1Dynamic Memory Architecture & Concepts

System Requirements

Parent Requirements:Windows Server 2008 R2 SP1Microsoft Hyper-V Server 2008 R2 SP1

Guest Requirements:Windows Server 2003, 2008 & 2008 R2

32-bit & 64-bit versions

Windows Vista and Windows 7Enterprise and Ultimate Editions only32-bit & 64-bit versions

Adding/Removing Memory

Adding MemoryEnlightened fashion

Synthetic Memory Driver (VSP/VSC Pair)

No hardware emulationLight weight

Removing MemoryWanted to remove memoryBallooning is more efficient

Messes up task manager in the guest OS

Dynamic Memory Architecture

Windows Kernel

Guest Applications

VMBus

Hypervisor

VMMS

OSMicrosoft Hyper-V

User Mode

KernelMode

Provided by:

VMBus

VMWP

VID

Memory Balancer

Memory Balancer Interface

GMOMemory Manager

DM VDEV/VSP

DM VSC

Dynamic Memory Components

Windows Kernel

Adding/Removing Memory

Active Memory additionMemory is added immediately when VM needs it

Passive Memory reclamationMemory is not removed when there is no immediate needUnutilized memory is collected every 5 minutes

Startup & Max

Startup: amount of memory to boot VMBIOS does not know about DMGuest OS may not know about DMDefault: 512MB

Max: don’t let the VM above this amountDefault: 64GB

Controlling Memory Availability

Availability is a conceptHow much memory does the VM have?How much memory does the VM want?The difference is the availability

Mechanisms in place to control Memory Availability through Buffer and Priority settings

Memory Buffer & Priority

Buffer: How much “free” memory should we try and keep in the VM?

Allows for responsiveness to bursty workloadsCan be used for file cache

“I like to configure my virtual machines so that they have ~20% free memory”

Priority: which VM gets the memory first1-10,000: default is 5,000The higher the priority, the higher the availability

Dynamic Memory in action

demo

System Impact

Changes to Root Reserve

Hyper-V has always had the concept of a reserve of memory that is kept for the parent partition

Host Reserve - Auto-calculated based on:If the machine SLAT capabilityTotal size of host RAMNUMA architecture

Host Reserve VM Overhead Memory utilized by VMs

VM Reserve

Changes to Root Reserve

DM allows VMs to push up against the reserve consistentlyNew behavior to better protect the parent partition from rampaging virtual machines

New registry key in placeAllows you to reserve static memory for the parent partition

May result in less memory being available for VMs

Guest Memory Stats

Physical Memory being used by the VM

Available Memory inside the VM

Total committed memory in the VM

Ballooned Memory2305 MB – 1720 MB = 585 MB

Changes to NUMA management…Wait – what is NUMA?Why do I care?How does this work today?

What is “NUMA”?A traditional computer:

Computer

CPU

CPU

CPU

CPU

BUS

Memory

VM VM VM VM

VM VM VM VM

What is “NUMA”?A NUMA computer:

Computer

CPU

CPU

CPU

CPU

BUS

Memory

VM VM VM VM

VM VM VM VM

Back Channel

Memory

BUS

Node 1 Node 2

Why do I care?VM memory should come from the “local NUMA node”

Computer

CPU

CPU

CPU

CPU

BUS

Memory

VM VM VM VM

VM VM VM VM

Back Channel

Memory

BUS

Node 1 Node 2

Good!

Why do I care?Ideally VM memory should come from the “local NUMA node”

Computer

CPU

CPU

CPU

CPU

BUS

Memory

VM VM VM VM

VM VM VM VM

Back Channel

Memory

BUS

Node 1 Node 2

Bad

How does this work today?Hyper-V tries to get all memory for a virtual machine from a single NUMA nodeWhen it cannot – the virtual machine “spans” NUMA nodesUsers can set preferred NUMA nodes for virtual machines in order to get the best distribution

Changes to NUMA managementDynamic memory can result in more virtual machines spanning NUMA nodes

A virtual machine might start all on one node – but added memory might come from another node

New option to disable NUMA node spanning

Disabling NUMA SpanningMakes the system behave like multiple small computers

Computer

CPU

CPU

CPU

CPU

BUS

Memory

VM VM VM VM

VM VM VM VM

Back Channel

Memory

BUS

Node 1 Node 2

What next?

Test the RC – and talk to us!Try Dynamic Memory out – for both server and desktop environments

Let us know how DM is working for youLet us know if you think we have something wrong

Q & A

Session Evaluations

Tell us what you think, and you could win!

All evaluations submitted are automatically entered into a daily prize draw* 

Sign-in to the Schedule Builder at http://europe.msteched.com/topic/list/   * Details of prize draw rules can be obtained from the Information Desk. 

 

© 2010 Microsoft Corporation. All rights reserved. Microsoft, Windows, Windows Vista and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries.The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to

be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.