Specifications

ManualsBrandsDell ManualsHome building and DecorPowerVault 200S

TEST REPORT

OCTOBER 2009

Dell PowerVault MD3000i high-availability testing on

Windows Server 2008 R2 and Hyper-V

KEY FINDINGS

• The highly-available Dell PowerVault

MD3000i solution easily supported our

various 200- and 500-user virtualized SMB

workloads. (See Figures 1 and 2.)

• The highly-available Dell PowerVault

MD3000i solution maintained outstanding

performance in both optimized and

degraded system states. (See Figures 3

and 4.)

• Our testing assumed all users were

running our four application workloads

simultaneously, meaning that the Dell

PowerVault MD3000i solution can easily

exceed our maximum of 500 simultaneous

users if running a similar workload under

real-world conditions.

Executive summary

Dell Inc. (Dell) commissioned Principled Technologies (PT)

to determine whether the Dell™ PowerVault™ MD3000i

iSCSI storage array in a high-availability configuration can

support small (200-user) and medium (500-user)

organizations simultaneously running a variety of virtualized

workloads in both an optimized and degraded system state.

We created a highly available solution with two Dell™

PowerEdge™ R710 servers, both running Windows

Server® 2008 R2 with Hyper-V as the underlying

hypervisor, using cluster shared volumes. Each Dell

PowerEdge R710 ran eight Windows Server 2008 R2 virtual

machines (VMs). The first Dell PowerEdge R710 ran a

Microsoft® Exchange 2007 virtual machine (which we ran

the Microsoft Load Generator test tool against), a file service

virtual machine (which we ran an Iometer workload against),

and six other background virtual machines that were active

but not under load. The second Dell PowerEdge R710 ran a

SQL Server® 2008 virtual machine (against which we ran the DVD Store test tool), an additional file service

virtual machine (against which we again ran an Iometer work load), and six other background VMs. We ran the

workloads simultaneously on each Dell PowerEdge R710 server with two different user counts: 200 and 500

simulated users. We assume these users are performing mail, database, and multiple file server operations

simultaneously. This means our testing exceeded real world conditions, where users would tend to exercise the

business infrastructure in a more sequential manner. We present more information on our test tools in the

Workload section of this report.

We installed a combination of

commonly-used Microsoft

applications on our background

virtual machines (those VMs that

were active but not under load),

including Microsoft SharePoint®,

Microsoft Internet Information

Server, Microsoft DHCP server,

and Microsoft Active Directory®.

We list all of the specific

applications we used in the

Workload section below.

Figure 1 presents the SQL Server

2008 performance results of the

Dell PowerVault MD3000i solution

in supporting a 200- and 500-user

count workload in an optimized

system state, where both servers

in our Windows 2008 R2 Hyper-V

cluster are operational and are

supporting virtual machines.

200

400

600

800

1,000

1,200

1,400

1,600

1,800

2,000

Operations per minute

Storage

Dell PowerVault MD3000i

SQL Server 2008 operations per minute

200 users

500 users

Figure 1: SQL Server 2008 Performance results of the Dell PowerVault MD3000i solution

in supporting 200- and 500-user count workloads in an optimized system state. Note:

We expect OPM to continue to rise linearly until the system is under maximum load. We

did not test maximum load for this study.

Summary of content (32 pages)

PAGE 1
TEST REPORT OCTOBER 2009 Dell PowerVault MD3000i high-availability testing on Windows Server 2008 R2 and Hyper-V Executive summary Dell Inc. (Dell) commissioned Principled Technologies (PT) to determine whether the Dell™ PowerVault™ MD3000i iSCSI storage array in a high-availability configuration can support small (200-user) and medium (500-user) organizations simultaneously running a variety of virtualized workloads in both an optimized and degraded system state.
PAGE 2
LoadGen latency (ms) 600 Dell PowerVault MD3000i Exchange Server 2007 latency 500 400 200 users 300 500 users 200 100 0 Storage Figure 2: Exchange Server 2007 Performance results of the Dell PowerVault MD3000i solution in supporting 200- and 500-user count workloads in an optimized system state. Lower numbers are better. The theoretical peak performance for our DVD Store workload is four operations per minute per user, for both the 200- and 500user workloads.
PAGE 3
We ran the LoadGen benchmark for 30 minutes on each Dell PowerEdge R710 server. For more details about LoadGen, see http://www.microsoft.com/downloads/details.aspx?FamilyId=DDEC1642-F6E3-4D66-A82F8D3062C6FA98&displaylang=en. DVD Store DVD Store Version 2 (DS2) is an open-source application with a back-end database component, a front-end Web application layer, and a driver layer that operates as the middle tier and actually executes the workload. The main DS2 metric is orders per minute, or OPM.
PAGE 4
• File server (active but not under load) Dell PowerEdge R710 #2 • Microsoft SQL Server 2008 (active and under DVD Store load) • File Server (active and under Iometer load) • Microsoft SharePoint (active but not under load) • File server (active but not under load) • Microsoft IIS (active but not under load) • Microsoft Systems Management Server (active but not under load) • Secondary Microsoft DHCP Server—different scope (active but not under load) • Microsoft SQL Server (active but not under load) We co
PAGE 5
Server Optimized SQL Server 2008 DVD Store Degraded SQL Server 2008 DVD Store Median OPM 200 users 760 758 Median OPM 500 users 1,878 1,868 Figure 3: Median DVD Store OPM results the second Dell PowerEdge R710 server and the Dell PowerVault MD3000i achieved during our measurement period under load for both optimized and degraded system states. Higher numbers are better.
PAGE 6
Iometer target workload We used Iometer to simulate a file server workload on both Dell PowerEdge R710 systems in order to add additional load to the storage array during our tests. We forced a target IOPS of .60 to .75 IOPS per user per file server and made sure the File Server IOPS stayed within the target range during our test runs.
PAGE 7
Figure 8: VM distribution on the Dell PowerVault MD3000i storage array and two Dell PowerEdge R710 servers in an optimized system state. Principled Technologies, Inc.
PAGE 8
Figure 9: VM distribution on the Dell PowerVault MD3000i storage array and a single Dell PowerEdge R710 server in a degraded system state. Principled Technologies, Inc.
PAGE 9
Appendix A: Test environment We used two Dell PowerEdge R710 servers to generate the workload and create demand on the storage. Figure 10 provides detailed configuration information for these servers.
PAGE 10
Server Speed (MHz) Speed in the system currently running @ (MHz) Timing/latency (tCL-tRCD-iRP-tRASmin) Number of RAM modules Chip organization Hard disk Vendor and model number Number of disks in system Size (GB) Buffer size (MB) RPM Type Operating system Name Build number File system Language Network card/subsystem Vendor and model number Type Optical drive Vendor and model number USB ports Number Type Dell PowerEdge R710 1,333 1,333 7-7-7-20 12 x 4 GB Double-sided Fujitsu Ltd. MBC2073RC 2 73.
PAGE 11
Appendix B: Dell PowerConnect 5448 switch Figure 11 presents the configuration of the two Dell PowerConnect 5448 switches we used. Component Details Switch Frame size Flow control iSCSI optimization Two Dell PowerConnect 5448 switches Jumbo frames enabled on all connected devices. Size set to 9,014 bytes Flow control set to Generate & Respond on all network interface cards iSCSI optimization disabled Figure 11: Dell PowerConnect 5448 switch configuration. Principled Technologies, Inc.
PAGE 12
Appendix C: Individual test beds Figure 12 shows diagrams of the components in each of the test beds and indicates their connections. Figure 12: The components in each test bed and their connections. Principled Technologies, Inc.
PAGE 13
Appendix D: Storage configuration Figure 13 shows the Dell PowerVault MD3000i, the primary storage hardware we used. Dell PowerVault iSCSI SAN The Dell PowerVault MD3000i has two controllers with two 1Gbps iSCSI ports per controller, one 100Mbps Ethernet management port per controller, and one SAS 4x port per controller. We connected the iSCSI switch to two iSCSI ports per controller and one management port per controller.
PAGE 14
Appendix E: Storage layout Figure 14 and shows the disk layout for each of the storage solutions we tested. We configured the MD3000i into four separate RAID 10 virtual disk groups. Disk group one contained the SQL Server database and the file server 1 dataset, Disk group two contained the Exchange Mailstore and the file server 2 dataset, Disk group three contained the SQL Server Logs and the Exchange Logs, and Disk Group four contained the operating system VHDs.
PAGE 15
Appendix F: Test methodology You can find Dell’s “PowerVault MD3000 and MD3000i Array Tuning Whitepaper” here: http://www.dell.com/us/en/enterprise/storage/pvaul_md3000i/pd.aspx?refid=pvaul_md3000i&s=biz&cs=555 We used the above whitepaper and Dell/Microsoft’s suggestions for Windows Server 2008 R2 with Hyper-V best practices, including the following: 1. Use virtual hard disks (VHDs) to allow the ability to use cluster-shared volumes (CSVs) and failover features in Windows Server 2008 R2.
PAGE 16
7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. Click Next. Accept the terms of the License Agreement, and click Next. Click Next. Click Next. Click Next. Click Install. Click Done. As the system restarts, press F2 to access the BIOS. Select Processor Settings, and press Enter. Change Virtualization Technology from Disabled to Enabled. Press Escape. Select Save Changes and Exit, and press Enter. In the Server Manager, click Roles.
PAGE 17
21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. Click OK. Click OK. Right-click the first client NIC, and select Properties. Select Internet Protocol Version 4 (TCP/IPv4), and click Properties. Select Use the following IP address and type 192.168.1.201 (on the Dell PowerEdge R710-1) and 192.168.1.
PAGE 18
70. Local Adapter: Set to Microsoft iSCSI Initiator. 71. Source IP: Select 192.168.130.201 (on the Dell PowerEdge R710-1) and 192.168.130.211 (on the Dell PowerEdge R710-2) 72. Target Portal: Select 192.168.130.102(iSCSI port 0-1) 73. Click OK. 74. Click OK. 75. Click the MD3000i, and click Log On. 76. Check Automatically restore this connection when the system boots. 77. Check the Enable multi-path checkbox, and click Advanced. 78. Local Adapter: Set to Microsoft iSCSI Initiator. 79. Source IP: Select 192.
PAGE 19
29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. Name the Virtual Disk SQL, and click Next. Click Finish. Click No. Name the Disk Group Logs and click Next. Select RAID level RAID 1/10. Set the number of Physical Disks to 2, and click Next. Click Yes. Name the Virtual Disk Logs, enter 128GB for the size, and click Next. Click Finish. Click Yes.
PAGE 20
82. Right-click the first unformatted disk, and select New Simple Volume. 83. Click Next. 84. Click Next. 85. Select F as the drive letter, and click Next. 86. Click Next. 87. Click Finish. 88. Right-click the first unformatted disk, and select New Simple Volume. 89. Click Next. 90. Click Next. 91. Select W as the drive letter, and click Next. 92. Right-click the first unformatted disk, and select New Simple Volume. 93. Click Next. 94. Click Next. 95. Select L as the drive letter, and click Next. 96.
PAGE 21
2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. Expand Hyper-V. Expand Hyper-V Manager. Click R710-1. Click Hyper-V Settings. Click Browse. Change the location to C:\ClusterStorage\Volume1, and click Select Folder. Click Virtual Machines. Click Browse. Change the location to C:\ClusterStorage\Volume1, and click Select Folder. Click OK. Right-click R710-1, and select NewÆVirtual Machine. Click Next.
PAGE 22
. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. Click Finish. Right-click the Windows Server 2008 R2 Virtual Machine, and select Start. Right-click the VM, and select Connect. Click Next. Click Install Now. Select Windows Server 2008 R2 Enterprise (Full Installation), and click Next. Check the I accept the license terms checkbox, and click Next. Click Custom (advanced). Click Next. Click OK. Enter your Administrator password, and confirm it. Shut down the virtual machine. Virtual Machine Setup 1.
PAGE 23
29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. Click the check box beside Web Server (IIS), and click Next. Accept all defaults by clicking Next. Click Next. Click Install. After installation finishes, review any warnings, and click Close. Click Run Setup.EXE. Click the link to Step 4: Install Microsoft Exchange Server 2007 SP1. Click Next to go past the introduction screen. Accept the license agreement, and click Next. Select No for error reporting, and click Next.
PAGE 24
31. Close the Exchange Management Console, and reboot the server. 32. Once the server successfully reboots, log in, and select StartÆAll ProgramsÆAdministrative ToolsÆActive Directory Users and Computers. 33. In the Active Directory Users and Computers window, select domain.company.com in the left pane, and open Users in the right pane. 34. Right-click a blank area in the Users folder, and select NewÆUser. 35.
PAGE 25
Installing and configuring Iometer 1. 2. 3. 4. 5. Download the Iometer package from www.iometer.org. Uncompress the files, and copy Iometer.exe, Dynamo.exe, and iometer.icf to c:\iometer. Open Iometer.exe. Verify that for Disk Targets, the # of Outstanding I/Os is set to the appropriate value. Under Results Display, make sure that Iometer has selected Start of Test, and set the update frequency to ∞. 6. Under Test Setup, set the Run Time to 30 minutes and the Ramp Up Time to 0 seconds. 7. Exit Iometer.
PAGE 26
3. Highlight Internet Protocol (TCP/IP), and click Properties. 4. Select the Use the following DNS server addresses radio button, and enter the IP address of the DNS server (192.168.1.250) in the Preferred DNS server field. Click OK. 5. Assign it an IP address of 192.168.1.25 and a subnet mask of 255.255.255.0 6. Right-click My Computer, and select Properties. 7. Under the Computer Name tab, click Change. 8. Assign a computer name of Client1 for the mail client. 9.
PAGE 27
Backing up the mail database On the test server, once you have set up LoadGen and created its initial mail database, you need to back up that database so you can have clean copies for each test. Use this process to back up that database: 1. Select StartÆAll ProgramsÆMicrosoft Exchange Server 2007ÆExchange Management Console in the mail server. 2. In the left pane, under Server Configuration, click Mailbox. 3. In the right pane, right-click Mailbox Database, and select Dismount Database from the menu. 4.
PAGE 28
On the command line for the ds2_create_orders.c module, we specified the size. The available options were S (small), M (medium), and L (large). We added the case W for the 250GB database. In the switch statement that sets the values for the variables max_cust_id and max_prod_id, we added cases that assigned them the proper values for the 250GB database size. We recompiled the ds2_create_orders.c module on Linux, following the instructions in the header comments.
PAGE 29
• • • We followed Microsoft’s recommendation of having 0.25 to 1 file per filegroup per core and we used 8 files per filegroup on our 24-core server. We did not use the DBCC PINTABLE command for the CATEGORIES and PRODUCTS tables, both because Microsoft recommends against this practice and because the commands do nothing in SQL Server 2008.
PAGE 30
The ds2connect routine in the ds2sqlserverfns.cs module defines sConnectionString. We used the following string; the changes we made appear in bold. string sConnectionString = “User ID=ds2User;Initial Catalog=“+dbname+”;Max Pool Size=200;Connection Timeout=120;Data Source=“ + Controller.target; Recompiling the ds2sqlserverdriver.exe executable We recompiled the ds2xdriver.cs and ds2sqlserverfns.cs module on Windows by following the instructions in the header comments.
PAGE 31
MOVE N'ds_misc4' TO N'F:\SQLData\DS2_1_20.ndf', N'E:\SQLData\DS2_1_21.ndf', MOVE N'ds_misc6' TO N'F:\SQLData\DS2_1_22.ndf', N'E:\SQLData\DS2_1_23.ndf', MOVE N'ds_misc8' TO N'F:\SQLData\DS2_1_24.ndf', N'E:\SQLData\DS2_1_25.ndf', MOVE N'orders2' TO N'F:\SQLData\DS2_1_26.ndf', N'E:\SQLData\DS2_1_27.ndf', MOVE N'orders4' TO N'F:\SQLData\DS2_1_28.ndf', N'E:\SQLData\DS2_1_29.ndf', MOVE N'orders6' TO N'F:\SQLData\DS2_1_30.ndf', N'E:\SQLData\DS2_1_31.ndf', MOVE N'orders8' TO N'F:\SQLData\DS2_1_32.
PAGE 32
About Principled Technologies We provide industry-leading technology assessment and fact-based marketing services. We bring to every assignment extensive experience with and expertise in all aspects of technology testing and analysis, from researching new technologies, to developing new methodologies, to testing with existing and new tools. When the assessment is complete, we know how to present the results to a broad range of target audiences.