File system fragmentation: Difference between revisions

Revision as of 07:07, 14 December 2006

Revision as of 07:52, 14 December 2006

Line 3:

In computing, '''[[file system]] [[fragmentation (computer)|fragmentation]]''', sometimes called '''file system aging''' is the inability of a file system to lay out related data sequentially (contiguously), an inherent phenomena in [[computer storage|storage]]-backed file systems that allow in-place modification of their contents. It is a special case of [[fragmentation (computer)#Data fragmentation|data fragmentation]].

File system fragmentation is projected to become more problematic with newer hardware due to the increasing disparity between sequential access speed and [[rotational delay]] (and to a lesser extent [[seek time]]), of consumer-grade [[hard disk]]s,<ref name=seagate-future>{{cite conference |author=Dr. Mark H. Kryder |publisher=[[Seagate Technology]] |date=2006-04-03 |title=Future Storage Technologies: A Look Beyond the Horizon |booktitle=Storage Networking World conference |url=http://www.snwusa.com/documents/presentations-s06/MarkKryder.pdf |accessdate=2006-12-14 }}</ref> which file systems are usually placed on. Thus, fragmentation is a important problem in recent file system research and design. The containment of fragmentation not only depends on the on-disk format of the file system, but also heavily on its implementation.<ref name=mcvoy-extent>{{cite conference |author=L. W. McVoy, S. R. Kleiman |date=1991 winter |title=Extent-like Performance from a UNIX File System |booktitle=Proceedings of [[USENIX]] winter '91 |pages=pages&nbsp;33&ndash;43 |___location=Dallas, Texas |publisher=[[Sun Microsystems, Inc.]] |url=http://www.cis.upenn.edu/~bcpierce/courses/dd/papers/mcvoy-extent.ps |accessdate=2006-12-14 }}</ref>

In simple file system [[benchmark (computing)|benchmark]]s, the fragmentation factor is often omitted, as realistic aging and fragmentation is difficult to model. Rather, for simplicity of comparison, file system benchmarks are often ran on empty file systems, and unsuprisingly, the results may vary heavily from real life access patterns.<ref name=workload-benchmarks>{{cite paper |author=Keith Arnold Smith |date=2001-01 |title=Workload-Specific File System Benchmarks |publisher=[[Harvard University]] |url=http://www.eecs.harvard.edu/vino/fs-perf/papers/keith_a_smith_thesis.pdf |accessdate=2006-12-14 }}</ref>

Line 20:

====Related file fragmentation====

Related file fragmentation refers to the lack of [[locality of reference]] of related files. Unlike the previous two types of fragmentation, related file fragmentation is a much more vague concept, as it heavily depends on the access pattern of relatedspecific filesapplications. This also makes objectively measuring or estimating it very difficult. However, arguably, it is the most critical type of fragmentation, as studies have found that the most frequently accessed files tend to be small.{{fact}}

==Techniques for mitigating fragmentation==

Several techniques have been developed to fight fragmentation. TheseThey can usually be classified into two categories: ''proactive'' and ''retroactive''. Due to the hard predictability of access patterns, these techniques are most often [[heuristic (computer science)|heuristic]] in nature, and may degrade performance under unexpected workloads.

====Proactive techniques====

Proactive techniques attempt to keep fragmentation at a minimum at the time data is being written on the disk. The simplest of such is, perhaps, appending data to an existing fragment in place where possible, instead of allocating new blocks to a new fragment.

Most today's file systems attempt to preallocate longer chunks for files that are actively appended to. This mainly avoids file fragmentation when several files are appended to concurrently, avoiding them from becoming excessively intertwined.<ref name=mcvoy-extent/>

A recent technique is [[allocate-on-flush]] in [[XFS]] and [[ZFS]], also called delayed allocation in [[reiser4]] and [[ext4]].