MIIS 2003 Capacity Planning Test Summary - Disk Performance

Updated: June 7, 2006

Applies To: Windows Server 2003 with SP1

Previous Sections in This Guide

This section presents different tests that evaluated disk performance. It describes the hardware platforms, data sets that were used, the test procedures, and the results.

The goals of the tests presented in this section are to arrive at a set of recommendations for selecting an optimal disk configuration for the MIIS 2003 server.

Test Description

To test disk performance, two rounds of tests were each performed on different disk configurations. During the first round, different groups of user objects, ranging from 10,000 to 500,000 objects, were first staged and then synchronized by using a server configured with only two internal disk drives. The operations per second were recorded to track performance.

For the second round, of tests the same groups of user objects were staged and synchronized. However, the server was reconfigured to use an external storage array that contained 10 spindles and was attached to the server by a SCSI interface. As with the first round of tests, the operations per second were recorded to track performance.

Expected Results

MIIS 2003 performance is directly related to Microsoft SQL Server performance because MIIS 2003 utilizes SQL Server for all its storage requirements. It is therefore expected that the performance of the MIIS 2003 environment would improve if the SQL Server performance is optimized through the use of multiple physical spindles and database file placement according to best practice recommendations. You can verify whether or not the SQL server has been optimized according to SQL server 2000 best practices by using the Best Practice Analyzer Tool found at the Microsoft Web site (http://go.microsoft.com/fwlink?linkid=55322).

Test Results Summary

The test results indicate that optimizing the SQL server that hosts the MIIS 2003 database will improve the performance of MIIS 2003. Both staging and synchronization operations were tested and both showed significant performance gains when the MIIS 2003 database was hosted on a SQL server configured for optimized disk I/O (in this case, by using an external SCSI disk array that contained 10 additional spindles).

Recommended Best Practices

Because MIIS 2003 relies on SQL Server 2000 for its database requirements, the overall MIIS solution performance will increase when the I/O subsystem is optimized. Thus is it is recommended that the I/O subsystem is optimized according to SQL Server 2000 best practices.

  • Place the database, log files, and TempDB file on different volumes – different physical spindles and different channels if possible.

  • More spindles improve performance.

  • Try to use RAID configurations that will yield the best performance while still providing for redundancy. For example, use RAID 0+1 split across various channels and as many drivers as possible.

Test Scenario

The scenario used during this testing was identical to the scenario that was configured during the previous hardware testing. The only difference is that during this specific round of testing, the first test was performed on a server configured to use its own internal storage. Subsequent tests were run on a server configured to use external SCSI attached storage units. This provided an easy way to monitor the impact that different I/O subsystem configurations could have on MIIS 2003 performance.

Server Hardware Configuration

For this test matrix, a single server utilizing two different disk configurations was used. As different disk configurations were tested the use of the same server helped eliminate other configuration factors such as memory speed, front side bus speed, and the storage controller cache that might have skewed the tests if completely different servers had been used.

The configuration of the server is described in the table below.

Table 17: Server hardware configuration

Server Designation Model Description

Q28

IBM xSeries 445

Quad Intel Xeon MP @ 2.8 GHz

No hyper-threading

4 GB System RAM

Internal LSI SCSI Controller

2x 34.6-GB Ultra 320 HDD (10k)

Windows Server 2003, Enterprise Edition (32-bit edition)

Storage Solution

IBM EXP 400

14 Disc SCSI Enclosure (10 slots used)

Storage Controller

IBM ServeRAID-6i

Used in the IBM xSeries 445 for connectivity to the storage enclosure.

Disk Configuration

During the first round of tests, the server was configured to only use internal storage, which in this case is a mirrored set of 36.4-GB SCSI U320 disk drives. Because this was the only physical volume, it hosted both the Microsoft SQL Server 2000 transaction log and the MIIS 2003 database.

During the second round of tests, the server was connected to an external IBM EXP 400 SCSI-attached storage unit. Both channels of the storage unit were used to increase I/O throughput and the storage unit, which hosted 10 additional hard disk drives, was configured as follows:

Table 18: Disk drive configuration

Drive Letter Function Description

G:

SQL Data Volume

4 x 18.2-GB 10,000 rpm (RAID 0+1)

F:

SQL Logs Volume

4 x 18.2-GB 10,000 rpm (RAID 0+1)

T:

SQL TempDB Volume

2 x 18.2-GB 10,000 rpm (RAID 1)

MIIS 2003 Configuration

MIIS 2003 with Service Pack 1 (SP1) was installed on the server being tested. The different roles of the servers in the test environment and the software installed are summarized in the following table.

Table 19: Server software configuration

Server Role Installed Software Description

MIIS 2003

MIIS 2003 SP1

Microsoft SQL Server 2000 SP3a

The server hosted the instance of MIIS 2003 being tested and the SQL database used by that instance of MIIS 2003. The text files for the text based management agents were also located on the MIIS 2003 server.

SQL data source

Microsoft SQL Server 2000 SP3a

This server hosted another SQL database that was used as a data source for some tests.

Active Directory data source

Microsoft Windows Server 2003, Enterprise Edition

This server hosted the instance of Active Directory that was used as a data source.

Management Agent Configuration

Three management agents were installed on MIIS 2003. The following table summarizes the configuration of each one.

Table 20: Management agent configuration

Management Agent Name Type Notes

ADMA

Active Directory

  • Located on a dedicated server.

  • At the beginning of each test, the directory is empty except for default administrative accounts created during setup.

SQLMA

SQL Server

  • Remote data source (dedicated SQL server).

  • Varying numbers of user objects based on the test cycle being performed (10,000, 50,000,100,000, 200,000, and 500,000).

  • Each user object has 25 attributes defined.

  • No multivalued attributes.

TXTMA

Text File

  • Local data source (text files used as the data source are located on the MIIS server)

  • Varying number of user objects based on the test cycle being performed (10,000, 50,000,100,000, 200,000, and 500,000).

  • Each user object has 25 attributes defined.

  • No multivalued attributes.

Run Profile Configuration

Separate run profiles were created for the staging and synchronization operations and both were configured as full run profiles rather than deltas. This was necessary because MIIS 2003 was reset after each test in order to begin each test with an empty connector space and metaverse. Because no objects existed prior to each test, full staging and synchronization operations were required rather than deltas.

During the testing process, all run profiles were executed sequentially. No concurrent operations were tested.

Rules Extensions

No rules extensions were used with any of the management agents. The only provisioning code used was the minimal amount required to create the user objects in Active Directory. This included the creation of Microsoft Exchange mailboxes for each user.

Test Results

The results of the disk I/O tests focus on the performance of the various platforms that were tested as measured by operations per second. This provides a simple way to compare overall performance.

Staging Performance

In this series of tests, the staging performance of the various platforms is compared during processing of the SQLMA and TXTMA run profiles, based on the number of operations that each platform completes per second.

Chart: Operations per second during staging

Table 21: Source data for Figure 12 (in operations/seconds)

Users Objects Tested Internal Storage EXP 400 SCSI Attached

10,000

91

278

50,000

109

243

10,0000

45

217

200,000

42

173

500,000

39

139

Observations
  • As expected with optimization on the I/O subsystem, the MIIS 2003 operations per second increase greatly.

  • There appears to be a linear performance on the external storage unit whereas the internal storage fluctuates somewhat. We were unable to determine the cause of the fluctuation.

Synchronization Performance

In this series of tests, the synchronization performance of the various platforms is measured based on the number of operations processed per second during all synchronization operations.

Chart: Operations per second during synch

Table 22: Source data for Figure 13 (in operations/seconds)

Users Objects Tested Internal Storage EXP 400 SCSI Attached

10,000

37

56

50,000

26

29

10,0000

9

27

200,000

8

16

500,000

5

10

Observations
  • A visible performance increase occurs when staging objects on the platform with an optimized I/O subsystem.

  • Again, as with the staging tests, there is an anomalous performance indicator on 50,000 objects. Once again we were unable to determine the cause of this.

External Factors and Other Considerations

Some additional hardware considerations regarding the tests discussed in this section:

  • The effect of different drive speeds was not taken into consideration. For example, multiple drive or array configurations were not tested. Instead, two data points where selected in an effort to prove that an optimized I/O subsystem will drastically increase the performance of an MIIS 2003 environment. This is due to the increases performance of the SQL server.

  • There is no single correct disk configuration. As long as you are seeing adequate performance based on the requirements of your specific environment, an alternate configuration is unnecessary.

Next

See Also

Community Additions

ADD
Show: