Diskeeper 2008 Enterprise Server review
Although disk drives were originally designed as random access devices, server operating systems include many techniques to improve their performance with sequential data. These work best when that data is written contiguously, but benefits can be lost when file fragments are scattered about the disk, and most applications will takea performance hit when their files are fragmented.
Disk-defragmentation software is fitted as standard on current versions of Windows, but finding the time to run it on a busy server, without upsetting the users, can be difficult. Diskeeper’s InvisiTasking feature attempts to counter this by defragmenting the data on the fly,using machine cycles to run the task continuously, although it can be suspended to allow for activities suchas scheduled backups. Furthermore, Diskeeper works on just one file at a time, reducing the load on the disk drive and keeping file locking to a minimum.
Microsoft’s NTFS filesystem is goodat resisting fragmentation, but its performance can suffer when the Master File Table (MFT) becomes fragmented. Diskeeper’s Frag Shield keeps the MFT defragmented and can monitor the page file performance. This feature has been upgraded and is now automatic. The new release also has the ability to handle the very large volume sizes available with the latest Windows server systems. The Terabyte Volume Engine is designed to work with volumes in excess of 60GB.
Microsoft’s Volume Shadow Copy Service has caused problems for previous releases of Diskeeper, but Diskeeper 2008 has a compatibility mode to combat this. It also includes the Intelligent File Access Acceleration Sequencing Technology (I-FAAST). This monitors disk activity to determine usage patterns and can move frequently used data files and applications to faster parts of the disk. Diskeeper claims this can improve file access and creation times by up to 80%.
We tested Diskeeper using a server equipped with twin Intel Xeon processors with 2GB of RAM running Windows Server 2003 and a workstation with 2GB of RAM and an AMD Athlon 64 dual-core processor 5000+, connected over a 10/100 network. According to Diskeeper, the server’s drive was already heavily fragmented. We ran a series of network copies between the server and workstation and some local file copies on the server itself. We then allowed Diskeeper to defragment the drive and reran the tests. The network copies showedno significant difference and network utilisation averaged 70%. In this case, the network was limiting performance. The tests on the server showed an average improvement of 10%.
A server has to do much more than push files across a network, and any server-based applications, including backups, should benefit from performance improvement. So not a resounding triumph, but Diskeeper did bring measurable improvement, so should interest anyone keen to keep a busy server running at its best.