NSTL Benchmarks Show NTFS Performance Improvements After Defragmentation

Although the benefits of hard drive defragmentation are well-known for the FAT file system, the benefits for NTFS have been surrounded in controversy, with different vendors providing different opinions. Even different groups within Microsoft have made conflicting statements about the benefits (or lack thereof) of performing disk defragmentation on NTFS drives. National Software Testing Lab (NSTL), in conjunction with Executive Software, has released the first independent tests showing the benefits of defragmentation for NTFS drives. NSTL tested four different applications: two desktop applications, Microsoft Excel and Outlook; and two server applications, Exchange 5.5 and SQL Server 7.0. To measure the applications' performance, NSTL used several script-based tests. The Excel test repeatedly opened and closed a set of four spreadsheets with formulas ranging in size from 4MB to 20MB. The Outlook and Exchange tests included opening 50 messages, moving 50 messages to a separate folder, and performing a text search of all messages in a folder. The Outlook test used a local Exchange Personal folder and the Exchange test used a server-based folder. The SQL Server tests consisted of importing a 400MB database plus two queries that opened multiple tables. NSTL conducted the desktop application tests on two basic Windows NT Workstation 4.0 system configurations and the server tests on two different NT Server 4.0 configurations. The NT Workstation tests used a Pentium II 266 with 96MB of RAM and a 2GB hard disk and a 400 MHz Pentium II with 288MB of RAM and a 4.2 GB hard disk. The company conducted the NT Server tests on a Pentium Pro 200 with 64MB of RAM and two 4GB SCSI drives and a dual Pentium Pro 200 with 128MB of RAM and five 4GB RAID 5 SCSI hard drives. NSTL used a proprietary fragmentation program to fragment the NTFS disks. Executive Software's Diskeeper disk defragmentation software measured the percentage of disk fragmentation and defragmented the disks. NSTL ran the tests before and after defragmentation. The test results showed significant performance benefits from disk defragmentation. The following table summarizes some of the NSTL findings.

Windows NT Workstation Tests

Configuration	Excel/Outlook Performance Improvement
Windows NT Workstation 266MHz	74.4%
Windows NT Workstation 400Mhz	80.6%

 

Windows NT Server Tests

Configuration	Exchange/SQL Server Performance Improvement
Windows NT Server 200MHz	19.6%
Windows NT Server 200MHz (dual)	56.1%

All tested system configurations showed significant performance improvements from NTFS defragmentation. Contrary to what you might expect, the faster system configurations benefited more from disk defragmentation than the lower-powered systems. You can find more information about the NSTL testing methodology and results in the white paper published by NSTL and Executive Software. You can find more information about NSTL at http://www.nstl.com. More information about Executive Software's Diskeeper is available at http://www.diskeeper.com.