Windows Server 2012 Networking Performance and ManagementAhmed TalatSenior Program Manager LeadMicrosoft Corporation

Prabu RambadranProduct Marketing ManagerMicrosoft Corporation

WSV304

Agenda

• Introduction

• Network management

• Managing your network performance

• Features and improvements

• Performance analysis and tools

Beyond Virtualization

Windows Server 2012 offers a dynamic, multi-tenant infrastructure that goes beyond virtualization to provide maximum flexibility for delivering and connecting to cloud services.

Modern Workstyle, Enabled

Windows Server 2012empowers IT to provide users with flexible access to data and applications from virtually anywhere on any device with a rich user experience, while simplifying management and helping maintain security, control and compliance.

The Power of Many Servers, the Simplicity of One

Windows Server 2012 offers excellent economics by integrating a highly available and easy to manage multi-server platform with breakthrough efficiency and ubiquitous automation.

Every App, Any Cloud

Windows Server 2012 is a broad, scalable and elastic server platform that gives you the flexibility to build and deploy applications and websites on-premises, in the cloud and in a hybrid environment, using a consistent set of tools and frameworks.

Windows Server 2012Cloud Optimize Your IT

Key TakeawaysBecome an Expert

• Understand some of the network management options available in Windows Server 2012

• Know when to use a certain network feature to meet your specific workload needs and goals

• Use the available tools to quickly diagnose performance problems and bottlenecks in your workloads

• Fine tune your systems and hardware to achieve optimum Networking Performance on Windows Server 2012

Network Management

• Resource Metering• IPAM• PowerShell (covered through out the deck)

title

Resource Metering

Resource Metering

Features

• Ability to create Resource Pools• Compatible with all Hyper-V

Operations• Data unaffected by Virtual Machine

Movement• Network Metering Port ACLs

Metrics

• Active CPU usage• Average memory use• Minimum memory use• Maximum memory use• Maximum disk allocation• Incoming network traffic• Outgoing network traffic

A two-tenant environment built with Hyper-V in Windows Server 2012

Resource Metering Continued …

Benefits• Track VM use• Aggregate data in a

Multi-tenant environment

• Build Chargeback solutions

• Easy to use PowerShell cmdlets

Basic model of Resource Metering

PowerShell CmdletsEnable resource Metering:Get-VM -ComputerName <Host Name> | Enable-VMResourceMetering

Get VM metering data:Get-VM -ComputerName <Host Name> -Name <VM Name> | Measure-VM

title

IP Address Management(IPAM)

IP Address Management today

.

.

.

.

.

.

.

.

• Automation• Rich feature set• Integration with own

and MS DHCP/DNS

High acquisition, support, and consulting costs

• Automation• High degree of

customization

Maintenance cost Relies on in-house

support model Expensive to add

new capabilities

• No CapEx investment

• Simple to use for small networks….at first

Labor intensive estimated (~$10 per address per annum)

Only performs address mgmt.

Inflexible and does not scale

3rd Party Solutions In-house tools Spreadsheets

Con

s P

ros

Windows Server 8 IPAM Basics

Consists of IPAM server and (RSAT) clientServer is in-box feature of Windows Server 8Available in all SKUs, except server coreClient runs on Windows 8 Client (RSAT) and Windows Server 8Agentless architecture

Supports MS DHCP and DNS Monitors and configures MS DNS and DHCP serversWS 2008, 2008R2 (and SPs), and WS 8

Manages both IPv4 and IPv6 addresses

Performance Features and Improvements

Let’s first cover some terminology• Processor. One physical processor, which can consist of

one or more nodes. A physical processor is the same as a package, a socket, or a CPU.

• Non-uniform memory architecture (NUMA) node. A set of logical processors and cache that are close to one another.

• Core. One processing unit, which can consist of one or more logical processors.

• Logical processor (LP). One logical computing engine from the perspective of the operating system, application, or driver. In effect, a logical processor is a thread.

• Kernel Group. A set of up to 64 logical processors.• Affinity. A preference indicated by a thread, process, or

interrupt for operation on a particular processor, node, or group.

Processor or Socket

NUMA Node

Core

LP LP

Kernel Group (KGroup)

Processor or SocketNUMA Node

Core

LP LP

Processor or SocketNUMA Node

Core

LP LP

Processor or SocketNUMA Node

Core

LP LP

Processor or SocketNUMA Node

Core

LP LP

Performance metrics definitions• Latency. The time required for an operation to complete. Lower is

better.

• Scalability. The ability to adapt to increasing demand on system resources. Higher is better.

• Throughput. The amount of data transferred or processed in a given time period. Higher is better.

• Path Length. The number of CPU cycles divided by the throughput. Lower is better.

• Variability / Jitter. Fluctuation in throughput and/or latency. Lower is better.

L

S

T

P

J

• IT Pro has received a heavy workload and CPU utilization is almost 100%. Unable to increase receive throughput

• Examples: Backup Server or FTP Server

• RSC can help by coalescing multiple inbound packets into a larger buffer and effectively reduce per packet CPU cost

• Preliminary data shows up to a 20% reduction in CPU cycles depending on the workload

• All inbox 10G network adapters support the feature and have it on by default

RSC: Multiple Inbound Packets

Coalesced to reduce CPU load

Networking FeaturesReceive Segment Coalescing (RSC)

T P

demo

Receive Segment Coalescing (RSC)

• Developer is writing an Application that needs to send and receive data at microsecond granularity.

• Examples: 3D image processing or stock trading application

• Registered I/O (RIO) delivers low latency by “pinning” the application memory and reduces CPU cost

• RIO improves predictability with message latency (jitter)

• Example APIs: RIORegisterBuffer and RIOReceive

Networking FeaturesRegistered I/O (RIO)

T PL S J

Low Latency Tunings and Considerations• IT Pro wants to fine tune their system

for low latency application and minimize jitter

• Best Practices include:• Disable C-states from BIOS• Set Power Policy to High Performance• Use powercfg utility

• Disable NIC Interrupt Moderation setting• Disable Flow Control• Disable Receive Side Scaling• Disable Receive Segment Coalescing (RSC)• Use thread affinity in Task Manager to

partition work by nodes• Check out http://

technet.microsoft.com/en-us/library/hh831415.aspx

Winsock Application (E.g. Using localhost as TCP destination)

AFD

TCP/IP

NDIS

kernel

user

Win08 R2 Loopback Path

Windows 2008 R2

Windows Server 2012 Loopback Path

New: Shorter Path

• Developer looking for an efficient and fast method to share data between processes on their Server.

• Examples: Database Servers and Business Applications like SAP

• Customer should use new IOCTL SIO_LOOPBACK_FAST_PATH socket option to take new shorter path

• Both sides of connection must set new option but can’t have SO_OOBINLINE, TCP_EXPEDITED_1122, or TCP_STDURG socket options

Networking FeaturesTCP Loopback Fast Path

T PL

demo

TCP Loopback fast path

• IT Pro has received a heavy workload that is CPU bound and wants to scale their workload across CPUs

• Examples: Web Server and File Server

• Use RSS to distribute the receive network traffic either across multiple processors or multiple NUMA nodes

• IT Pros should use new Power Shell cmdlets to control RSS behavior based on their workload characteristics

PS C:\cim> Set-NetAdapterRss -? PS C:\cim> Set-NetAdapterRss -InterfaceDescription <String[]> -Profile <UInt32> -Confirm

DynamicStatic

NUMA aware

NonNUMA

Profile 1

Profile 2Profile 3(default)

Profile 0 (like WS08

R2)

Networking FeaturesReceive Side Scaling (RSS)

TS

NIC with 8 RSS queues

Node 0 Node 1 Node 2 Node 3

qu

eu

es

2

2

3

3

1

1

0

0

Incoming Packets

New RSS profiles improve scalability on multi-node servers by distributing TCP/UDP receive traffic across nodes

Networking FeaturesReceive Side Scaling (Continued)

Th

roughp

ut

2 Nodes (20LP)

4 Nodes (40LP)

WS08R2Win2012

Web Scalability ImprovementsSingle Process

Hardware: 40LPs, 4 NUMA nodes , 10GigE, 128GB RAMWorkload: mixed Web scenario

Note: Measurements reflect release defaults

50%

201%

Multiple Processes (2 and 4)

2 Nodes (20LP)

4 Nodes (40LP)

WS08R2Win2012

Th

roughp

ut

43%

115%

Overall Web Server Scalability Improvements• New K-Group awareness in the IIS thread pool• Partitioning network I/O processing by K-Groups when using multiple

NICs• RSS NUMA profiles distribute network I/O processing across NUMA nodes

• An IT Pro wants to run more VMs on a physical machine but can’t because the incoming packet processing is saturating a limited set of CPUs on the physical host

• Examples: Private Cloud Deployments • Use VMQ to dynamically distribute the receive network

traffic across multiple VMs based on destination and network load

• Most inbox 10Gig Ethernet adapters support the feature which is enabled by default

Networking FeaturesDynamic Virtual Machine Queues (VMQ)

TS

Network IO path without VMQ Network IO path with VMQ

• Windows Server 2008 R2: Offloading routing and filtering of network packets to the network card (enabled by hardware-based receive queues) to reduce host overhead

• New in Windows Server 2012: Dynamically distributing incoming network traffic processing to host processors (based on processor utilization and network load)

Networking FeaturesDynamic VMQ (Continued)

Series1

1.8x scaling from 40VMs on 80 LPs to 80VMs on 160 LPs with dynamic VMQ

• Hardware: 160LPs, 8 NUMA nodes, 10Gigabit NIC, 512GB RAM• Workload: 2VPs/VM mixed Web

scenario• Scenarios: Web Hosters

80LPs

160LPs

Th

roughp

ut

Larger is better

Windows Server 2012 ImprovementsHyper-V Host Scaling

title

Performance Analysis and Tools

demo Identifying performance bottlenecks using SPA 3.0

Can Turn On/Off

Can Modify Thresholds

Original Setting is derived from system tuning, empirical

evidence

• Zero install on servers

• Side by Side Comparison Report

• Historic Performance tracking and trending

• Thresholds can be modified

• Guide Performance optimizations via recommendations and next steps

EXAM

PLE R

ULE

Performance Analysis and ToolsServer Performance Advisor (SPA) 3.0

• Microsoft Winsock BSP• Datagrams Drops or Rejected

• NUMA Node Memory• Per-NUMA memory lists• Overall system memory info per

Node• Network Adapter• Processor Performance

• Power related counters• Physical Network Interface

Activity• RDMA Activity

Performance Analysis and ToolsNew Performance Monitor Counters

• Get more information on hardware settings• Get-NetAdapter W*n• Get-NetAdapter L*n* -IncludeHidden (e.g. SSTP, PPTP)• NetAdapter.Get-NetAdapterIPsecOffload• NetAdapter.Get-NetAdapterLso• NetAdapter.Get-NetAdapterRss• NetAdapter.Get-NetAdapterRsc• NetAdapter.Get-NetAdapterVmq | Where-Object –FilterScript {$_.Enabled}• NetAdapter.Get-NetAdapterSrioVf –Name “Ethernet 1”

• Set specific properties on the hardware• NetAdapter.Set-NetAdapterRss –Name “Ethernet 1” –Profile NUMAStatic• NetAdapter.Set-NetAdapterSriov –Name “Ethernet 1” –NumVFs 31 –VPorts

64

EVERYTHING is configurable through PowerShell

Performance Analysis and ToolsPowerShell commands

• Enabling Performance features on the hardware• NetAdapter.Enable-NetAdapterEncapsulatedPacketTaskOffload

MyAdapter• NetAdapter.Enable-NetAdapterIPsecOffload MyAdapter• NetAdapter.Enable-NetAdapterLso MyAdapter –IPv4• NetAdapter.Enable-NetAdapterRdma MyAdapter• NetAdapter.Enable-NetAdapterRsc MyAdapter

• Get-Help <cmdlet name> -Detailed• Get-Help <cmdlet name> -Examples• Get-Help <cmdlet name> -Full

Performance Analysis and ToolsPowerShell commands (Continued)

SummaryWhen do you use each feature?

Performance Metric

Loopback Fast Path

Registered I/O (RIO)

Large Send Offload (LSO)

Receive Segmentation Offload (RSC)

Receive Side Scaling (RSS)

Virtual Machine Queues (VMQ)

Remote DMA (RDMA)

Single Root I/O Virtual(SR-IOV)

Lower End-to-End Latency

X X X X

Higher Scalability X X X

Higher Throughput X X X X X X X X

Lower Path Length X X X X X X

Lower Variability X

Conclusion

• Performance improvements across a number of key Windows Server 2012 roles

• Out of box performance improvements for LAN and WAN scenarios

• Available tools for networking performance analysis and optimizations

• Quick overview of network management options in Windows Server 2012

Related Content

VIR303 – An Overview of Hyper-V Networking in Windows Server 2012

VIR305 – Hyper-V Network Virtualization for Scalable Multi-Tenancy in Windows

MGT315 - Network Management in Microsoft System Center 2012 SP1

WSV314 – Windows Server 2012 NIC Teaming and Multichannel Solutions

Visit networking booth at the partner pavillion to learn more

Related Content

WSV321: Windows PowerShell Crash CourseWSV307: Windows Server 2012 IP Address ManagementWSV314: Windows Server 2012 NIC Teaming and Multichannel SolutionsVIR303: An Overview of Hyper-V Networking in Windows Server 2012VIR305: Hyper-V Network Virtualization for Scalable Multi-Tenancy in Windows VIR307: Get Hands-on with the New Hyper-V Extensible Switch in Windows Server 2012WSV14 – HOL: Managing Your Network Infrastructure with IP Address ManagementWSV24 – HOL: Managing Network Infrastructure with Windows Server 2012

SIA, WSV, and VIR Track Resources

Talk to our Experts at the TLC

#TE(sessioncode)

DOWNLOAD Windows Server 2012 Release Candidate

microsoft.com/windowsserverHands-On Labs

DOWNLOAD Windows Azure

Windowsazure.com/teched

Resources

Connect. Share. Discuss.

http://northamerica.msteched.com

Learning

Microsoft Certification & Training Resources

www.microsoft.com/learning

TechNet

Resources for IT Professionals

http://microsoft.com/technet

Resources for Developers

http://microsoft.com/msdn

Complete an evaluation on CommNet and enter to win!

MS Tag

Scan the Tagto evaluate thissession now onmyTechEd Mobile

© 2012 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.