How to solve file fragmentation problems that plague server performance

The file system's own mechanism causes each computer to suffer from file fragmentation. The system and file sharing functions of the server are subject to numerous user accesses, which causes file fragmentation problems to be particularly serious. The most common consequence is slow network speed and booting. Long time and virus scan time are particularly long. Is the current hot technology virtualization and SSDs effective in addressing the impact of fragmentation on storage system server performance?

The Impact of File Fragments on Virtualization Many organizations have begun to take advantage of virtualization for server consolidation and data center carbon emissions. But the advantages of this future-oriented technology are not limited to this. The key challenge facing virtualization performance is the processing of file fragments.

File fragmentation is a major performance challenge that has existed in non-virtualized environments and has remained unresolved for many years. But the fragmentation problem caused by virtualization is more serious, and the solution required is unprecedented. Virtual machines can take advantage of hard disk partitioning, making the entire disk system look like a virtual machine. However, under the virtualization layer, the hardware often stores the files generated by the system, and uses the disk storage system and fragment files of all partitions of the entire hard disk.

Virtual machines have their own input/output requests that are passed through the host system. So each file request will generate a lot of input/output requests -- to a minimum, the client system will generate a request, and then the host system will generate another request. But in the conventional smashing process, especially when the virtual machine disk activity is frequent, the file will be shredded into hundreds of file fragments. It is conceivable how many input/output requests generated by each fragment of each requested file and file will evolve into frantic actions. How terrible the impact on performance will be!

For virtual machines, regular disk defragmentation is critical, and this takes advantage of disk fragmentation. Basic disk defragmenters, even pre-configured disk defragmenters, can't keep up with the smashing speed of virtualization. The most viable and only solution available today is the continuous background shredding solution. This solution handles those running in the background or idle; this is a very positive effect on the performance of the defragmenter and always maximizes performance.

Shards are easy to get

SSDs have unique advantages for traditional hard drives, but a recent test from a foreign media site may cause Intel's famous X25-M SSD to lose some Attractive. A recent test from the PCPerspective website shows that Intel's wear-levelling features can cause large-scale disk fragmentation problems, and even more serious, these problems can hardly be fixed. It seems that Intel's X25-M SSD product seems to have some problems.

This website called PCPerspective has published a very detailed long-term performance test report on SSDs. During the test, the testers found that some of the technologies that have achieved good results on conventional hard drives are in SSDs. The above does not apply. The article lists two typical examples of sector remapping and wear leveling algorithms. The role of the two on conventional hard disks is to accelerate the performance of the SSD and extend the life of the hard disk, but tests have shown that Both technologies have greatly reduced the performance of SSDs after the SSDs have been fragmented. To make matters worse, regular disk defragmentation software can not only solve this problem but it will aggravate the situation. Considering that this disk fragmentation is generated at the file system level, Intel also has SSD engineers who advise users not to use any Disk defragmentation software to do anything with the SSD drive. The only way to re-clear drive leaf mapping is to empty the entire disk again.