HP: Performance of SharePoint 2010 and SQL 2012 with HP Gen8 – In the lab and in the Real World

Chris Daly – Sr. Engineer / Alliances, Performance, and Solutions (APS)

Mike Linster – Collaboration Lead / U.S. Network Consulting Delivery

SPC160

ProLiant Gen 8: What’s New

Next generation ProLiant with Active Insight architecture

Operational efficiency

Applicationacceleration &

availabilityAlways On Support

• Setup in seconds

• One touch update

• Continuous health and diagnostics

• Total control anytime, anywhere

• CPU Performance: 1.5x – 1.75x performance increase

• Storage: 2x faster, 50% more density, 95% faster rebuilds

• Foundation Care + Collaborative Support Services

• Proactive Care Services

• Insight Online; Insight Remote Support

Data Center Optimization

• Datacenter 3D energy optimization

• Self-identifying and self-locating system inventory

HP LoadRunner• Test a broad range of

applications, including the latest RIAs, Web 2.0 technologies, mobile applications, ERP/CRM applications, as well as legacy technologies

• Support testing on traditional platforms, as well as new platforms such as cloud, hybrid environments, and mobile

• Identify and reduce performance bottlenecks before deployment to avoid performance issues in production

• Obtain an accurate picture of end-to-end system performance before going live, and verify that new or upgraded applications meet specified performance requirements

SharePoint 2010 and HP LoadRunner

How to Build a Script for

SharePoint• Record the task you want to

automate.• Parameterize the Variables

• The Magic 3• EventValidation• ViewState• RequestDigest

• Build in logic• Build in Transaction Markers• Build in wait times

Sharepoint 2010 Enterprise Collab Workload

5 Primary Actions

1. Document Center30%

2. Team Sites20%

3. Portal20%

4. Mysite10%

5. Search20%

• Document center

• Document check out / Upload / check in

• Read / Create / Delete Task

• Team Sites – 25 different Team sites

• Upload documents

• Read / Create / Delete Calendar entries

• Read / Reply Thread Discussions

• Portal

• Read Events

• Read Announcements

• Respond to Survey

• Mysite host

• Upload Document to Personal Documents

• Search

• Searching / Downloading documents

Testing configuration

• Web Front End• HP ProLiant DL380p Gen8

• (2) Intel® Xeon® E5-2690 CPUs• 192 GBs of RAM• (2) HP 300GB 6G SAS 15K rpm SFF Enterprise

drives

• SQL Server 2012• HP ProLiant DL380p Gen8

• (2) Intel® Xeon® E5-2690 CPUs• 192 GBs of RAM• (2) HP 300GB 6G SAS 15K rpm SFF Enterprise

drives

ProCurve Switch 2900-48G

HP ProLiant DL380p Gen8 Server

SQL Server 2012

HP ProLiant DL380p Gen8 Server

Web Front End

HP P4500 G2 7.2TB SAS Storage System

Front End CPU Performance @ 250 Users

CPU Utilization

28.5% (Peaks to 47%)

Memory Usage

18GBs of RAM

IOPS

29.5 (Peaks to 348)

Back End CPU Performance @ 250 Users

CPU Utilization

3.9% (Peaks to 8.5%)

Memory Usage

44GBs of RAM

IOPS

1037 (Peaks to 2580)

Network Consumption @ 250 Users

Client to IIS

470 Mb/s(648 Mb/s peak)

IIS to SQL

410 Mb/s(734 Mb/s peak)

Virtualizing with Windows Server 2012• Up to 32 vCPUs per VM• Up to 512GB of RAM per VM• VM disks up to 16TB with the new VHDX format• Multiple concurrent Live Migrations and Storage Live

Migrations• Hyper-V Replica• Extensible virtual switch• 63 node clusters• Support for native NIC teaming

Testing Configuration• HP ProLiant DL380p

Gen8• (2) Intel® Xeon® E5-2690 CPUs• 192 GBs of RAM• (2) HP 300GB 6G SAS 15K rpm

SFF Enterprise drives

• VM1 – Web Front End• 24 vCPUs• 32GBs of RAM• XXGB VHD host on internal drives

• VM2 – SQL Server 2012• 8 vCPUs• 64GBs of RAM• 1TB VHD host on HP P4500

Storage System

ProCurve Switch 2900-48G

HP ProLiant DL380p Gen8 Server

HP P4500 G2 7.2TB SAS Storage System

VM1 – WFE24 vCPUs

32GBs RAM

VM2 – SQL8 vCPUs

64GBs RAM

Front End CPU Performance @ 250 Users

CPU Utilization

36% (Peaks to 78.5%)

Memory Usage

12GBs of RAM

IOPS

60 (Peaks to 228)

Back End CPU Performance @ 250 Users

CPU Utilization

14.7% (Peaks to 28%)

Memory Usage

42GBs of RAM

IOPS

1138 (Peaks to 2797)

Network Consumption @ 250 Users

Client to IIS

465 Mb/s(688 Mb/s peak)

IIS to SQL

408 Mb/s(677 Mb/s peak)

Real World Situations

Canadian Telecommunications Company

Workloads:WCMCollaboration ECM

Average RPS:75Peak RPS:150

Average Users:1,000Peak Users:3,000

Original Architecture

Testing Performed

Testing Tools:HP Load CenterHP Open View MonitoringPerfMonNetMon

Types of Tests:BrowserWordWorkspaceDAVPowerPoint

Testing Results for the Home PageRemember the expected concurrent users was set at 3,000 users at peak time

750 Virtual Users:Minimum: .86 SecMaximum:119.83 Average: 25.30

250 Virtual Users:Minimum: .73 SecMaximum:41.36 Average: 7.612

1500 Virtual Users:Minimum: .77 SecMaximum:259.63 Average: 35.26

Updated Architecture

Testing ResultsRemember the expected concurrent users was set at 3,000 users at peak time

750 Virtual Users:Minimum: .71 SecMaximum:2.65 Average: 2.95

250 Virtual Users:Minimum: .73 SecMaximum:2.29 Average: 2.49

1500 Virtual Users:Minimum: .71 SecMaximum:2.3 Average: 2.25

Issues Revealed

HOST CPU Contention

NUMA Boundaries

Clock Synchronization

Load Balancer Configuration

IOPS under estimated

Memory Ballooning

Package Company

Workloads:WCMCollaboration SocialCustom Apps

Average RPS:UnknownPeak RPS:Unknown

Average Users:10,000Peak Users:30,000

Original Architecture

Testing Performed

Testing Tools:User Tests

Types of Tests:Browser

Testing Results for the Home Page

Home Page5 users the site would more often than not time out

My Site host:5 users over 45 seconds

Project / CommunityExtended to 30 to 45 seconds

Updated Architecture

Testing ResultsCurrent Beta Iteration

Home Page1,500 users under the 5 second threshold

My Site host:1,500 users under 5 seconds

Project / Community1,500 users under the 5 second threshold

Issues Revealed

HOST CPU Contention

NUMA Boundaries

Clock Synchronization

Load Balancer Configuration

IOPS under estimated

Memory Ballooning

Non Optimized JavaScript

Stored Procedures

Clock Synchronization

Lack of Caching Lock Contention within SQL

SQL Configuration

DB Dynamic expansion

Lessons Learned

Virtual Environments

Mix roles on Hosts

VHD should not be stored on host storage

Host memory ballooning

Over Commitment

Clock Synchronization with the Host is not always best

NUMA Boundaries need to be understood

All Environments

Custom UI needs to be well controlled and managed

Monitoring, Monitoring, Monitoring

Cache should not be overlooked at all levels

SQL TuningLocation of Servers

Pre-release performance testing

© Copyright 2012 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.33

Don’t Forget to Get Your Card Stamped!Attend these sessions

Monday, Nov12 2:25 – 2:40 pmExpo Hall Partner TheaterA framework for Microsoft SharePoint operations using HP Enterprise Cloud Services

Tuesday, Nov13 2:15 – 2:30 pmExpo Hall Partner TheaterAn introduction to HP Private Cloud SharePoint Services

Tuesday, Nov13 3:15 - 4:30 pmIslander Ballroom CHPerformance testing results of SharePoint 2010 and Microsoft SQL 2012 with HP Gen8

Wednesday, Nov14 11:45 am-1:45 pmMariners ABOvercome the top 5 challenges with supporting and optimizing your SharePoint investment

After the event

• Contact your sales rep

• Visit the website / Facebook /Twitter at: <insert URL here>

• Download the whitepaper at: <insert URL here>

GOT QUESTIONS? WE’VE GOT ANSWERS! MEET OUR SHAREPOINT EXPERTS AT BOOTH #307

Evaluate this session now on MySPC using your laptop or mobile device: http://myspc.sharepointconference.com

MySPC

© 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.