|
With continued expansion of organizational storage allocation requirements, the time required for file-based data backup also continues to increase yet the time allotted for this data backup remains constant, exposing a critical issue. Though backup software vendors continue making advancements in technology to speed up the process, disk fragmentation will continue to remain an unresolved issue that must be addressed on the file system level to solve this bottleneck.
In business environments, backing up local and network data is commonly done with both tape devices and disk devices. As data backup involves file access, fragmentation of data files can have a profound impact on the length of time a backup procedure may take.
Diskeeper Corporation, an industry-leading developer of systems management tools, performed a series of rigorous tests to determine the effects of fragmentation on backup operations and the benefits of defragmentation.
In conclusion, Diskeeper Corporation has found that the time required to back up data from a typical hard drive volume to these backup devices can be decreased and the backup data transfer rate increased by defragmenting the volume prior to backup. Furthermore, directory consolidation (defragmenting and consolidating directory files) has a direct additional benefit in both of these areas. The study showed improvements in backup times up to 69% (decrease in total back up time)-less than one half of the time for the fragmented case. Similarly, backup data transfer rates increased up to 69% over that for the fragmented state-nearly 1.7 as fast.
Test Methodology:
Overview
A set of trials was conducted to determine the benefit that file and directory defragmentation have on file backup performance for a server.
Setup
Hardware:
1. Local Test system running Windows 2003 Server Enterprise with Service Pack 1:
- Intel D925CV2 motherboard (BIOS date 10/12/04) with 3.6 GHz Pentium 4 CPU
- 2GB RAM
- 200GB Seagate ST3200826AS SATA hard drive
2. Network remote backup system running Windows 2003 Server Enterprise with Service Pack 1:
- Dual 1GHz Pentium 3 CPUs on an Asus CUR-DLS motherboard, BIOS revision 1009
- 1GB ECC memory (4x256MB DIMMs)
- Seagate 36GB 10,000 RPM Cheetah SCSI hard drive
- Mylex Extreme RAID 2000 SCSI-160 controller with 32MB cache memory, BIOS PLUS version 6.01-39.
- Intel PRO/1000 MT 1Gbit/s network interface card
- Adaptec 2940-UW SCSI controller (for local tape backup system)
- Exabyte EZ-17 Mammoth-2 tape external backup system (connected to outboard 69-pin connector on Adaptec 2940 via 6 foot shielded cable
3. Network remote backup system running Windows 2000 Server with Service Pack 4:
- Dual 1GHz Pentium 3 CPUs on a DL380 motherboard, BIOS revision 2.10 (revised 12/18/02)
- 1GB DDR (2 x 512MB DIMMs)
- Seagate ST380011A IDE 80GB hard drive
Backup Software Used:
1. Symantec Backup Exec 10d (Build 5629)
2. Computer Associates BrightStor ARCServe Backup release 11.5 (Build 3884)
Defragmentation Software Used: Diskeeper 10.0.
Test methodology:
1. Create an image of the operating system partition on the test server.
2. Create and image a fragmented 20 gigabyte test volume containing approximately 100,000 files, about 60% data fragmentation and about 40% free space. (See Figure 1 in the Appendix.)
3. Install and set up one of the backup programs to test with (Backup Exec or
ARCServe Backup).
4. Back up the test volume to a local disk backup device (created with the backup program), and record the backup time and data transfer rate for the fragmented volume.
5. Install the backup software on another system on the same network.
6. Back up the test volume over the network to a disk backup device (created with the backup software), and record the backup time and data transfer rate for the fragmented volume.
7. Back up the test volume over the network to a tape backup device, and record the backup time and data transfer rate over the fragmented volume.1
8. Defragment the test volume using Diskeeper Server Enterprise Edition (using different defragmentation methods to see what the results were).
9. Repeat steps 4-8 with the defragmented test volume.
10. Restore the image of the fragmented test volume.
11. Repeat steps 3-9 for the next backup program.
|
