GANGER, G. R., AND KAASHOEK, M. F. Embedded inodes and explicit grouping: exploiting disk bandwidth for small files. In Proc. of the 1997.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....blocks; and (4) the management of the controller cache cannot be excessively complicated to avoid overheads. These differences mean that techniques proposed for other caches rarely apply to controller caches. Perhaps the most closely related operating system level technique is explicit grouping [12]. This technique groups small files that belong to the same directory into consecutive disk blocks, so that read aheads fetch useful data when multiple files in the directory have to be accessed. Explicit grouping shares the same goal as FOR, but requires that a meaningful grouping of files be ....

Gregory R. Ganger and M. Frans Kaashoek. Embedded Inodes and Explicit Grouping: Exploiting Disk Bandwidth for Small Files. In Proceedings of the USENIX Annual Technical Conference, pages 1--17, January 1997.

....file system which works with the same commonly used operating systems as their research, but is significantly faster because it is not constrained by the semantics of general purpose irreplacable data. 2. 2 Constituent Mechanisms The primary source for Damelo s explicit grouping idea is [7]. The authors aggressively pursue adjacency of small objects rather than just locality. Like Damelo, they espouse the idea of large blocks: a 64kb access takes less than twice as long as an 8kb access. Because they o#er conventional file system integrity as part of their Co locating Fast File ....

G. Ganger and M. Kaashoek. Embedded inodes and explicit grouping: Exploiting disk bandwidth for small files, 1997.

....dynamic disk accesses, improving performance by enabling the cache for the former and disabling it for the latter. The Markov models were created per le. In [MADH96] disk traces were collected dynamically and used to adaptively tune le system parameters in a parallel computing environment. [GANG97] improved performance by placing single block and other small les adjacent to their inodes, and combined inodes physically with their directories. PATT91] suggested that ap5 plications could give the operating system hints about when to prefetch. Many modi cations to existing le systems as ....

G. R. Ganger and M. F. Kaashoek, Embedded Inodes and Explicit Groupings: Exploiting Disk Bandwidth for Small Files, in Proceedings of the USENIX Technical Conference, pages 1-17, January 1997.

....that files in the same directory are likely to be accessed together. This idea, along with the concept of immediate files (files that are short enough that it makes sense to embed them in the same disk page as their metadata [11] has been extended to create variants of FFS such as C FFS [5], which co locates inode and directory information to improve directory lookups and uses explicit grouping to allow groups of small files in the same directory to be accessed entirely sequentially. These heuristics, while effective in many cases, are limited because they are static and do not ....

Gregory R. Ganger and M. Frans Kaashoek. Embedded inodes and explicit grouping: Exploiting disk bandwidth for small files. In USENIX Annual Technical Conference, pages 1-17, 1997.

....all inter client communication. Files are fragmented, and each fragment (blocks of size 4 KB) is encrypted independently with its own file block (3DES) key. This 3DES key is kept in the fragment s lock A similar problem was encountered in the context of storing inodes and small files together [16]. box together with the length of the fragment. The hashes of all the fragments are arranged in a Merkle hash tree, and the root signed (1024 bit RSA) with the file sign key. The leaves of the tree contain the lockbox of the corresponding fragment. The tree is kept in a shadow file, on the ....

G. Ganger and M. Kaashoek. Embedded inodes and explicit grouping: Exploiting disk bandwidth for small files. In USENIX Tech. Conf., pages 1--17, 1997.

No context found.

G. R. Ganger and M. F. Kaashoek. Embedded inodes and explicit grouping: exploiting disk bandwidth for small files. USENIX Annual Technical Conference, pages 1--17, January 1997.

No context found.

Gregory R. Ganger and M. Frans Kaashoek. Embedded inodes and explicit grouping: exploiting disk bandwidth for small files. USENIX Annual Technical Conference (Anaheim, CA), pages 1--17, January 1997.

No context found.

Gregory R. Ganger and M. Frans Kaashoek. Embedded inodes and explicit grouping: exploiting disk bandwidth for small files. USENIX Annual Technical Conference (Anaheim, CA), pages 1--17, January 1997.

No context found.

Gregory R. Ganger and M. Frans Kaashoek. Embedded inodes and explicit grouping: exploiting disk bandwidth for small files. USENIX Annual Technical Conference (Anaheim, CA), pages 1--17, January 1997.

No context found.

Gregory R. Ganger and M. Frans Kaashoek. Embedded Inodes and Explicit Grouping: Exploiting Disk Bandwidth for Small Files. In USENIX Annual Technical Conference, pages 1--17, 1997.

....File based heuristics use information about file inter and intrarelationships to co locate related blocks. For example, most file systems try to allocate blocks of a file sequentially. C FFS allocates adjacently the data blocks that belong to multiple small files named by the same directory [5]. Hummingbird and Cheetah perform similar grouping for related web objects (e.g. an HTML document and its embedded images) 8, 23] Other system management policies: Relevant research has also been done on storage system policies, such as caching and prefetching [2, 7, 16] Most notably, ....

G. R. Ganger and M. F. Kaashoek. Embedded inodes and explicit grouping: exploiting disk bandwidth for small files. USENIX Annual Technical Conference, pages 1--17, January 1997.

.... in proximate cylinders, objects that are likely to be accessed around the same time (e.g. a file s metadata and data) 35] More recent work proposes embedding file metadata in directories, and storing the data of small files from the same directory in adjacent blocks that are accessed as a unit [13]. A second approach is to rearrange data that is already on disk. Researchers have explored reducing seek times by rearranging on disk data such that frequently accessed data is near the middle of the disk [66, 55, 44, 1] This research exploits the observation that, in general, there is a small ....

G.R. Ganger and M.F. Kaashoek. Embedded inodes and explicit grouping: Exploiting disk bandwidth for small files. In Proceedings of the USENIX Winter 1997.

....24, 33] Even for general file system operation, allocation of very large sequential regions competes with space management robustness [25] and very large accesses may put deep prefetching ahead of foreground requests. Also, large requests can be used for small files by grouping their contents [14, 15, 17, 32, 33], but larger requests require grouping more files with weaker inter relationships. These examples all indicate that achieving higher disk efficiency with smaller request sizes would be valuable. This paper describes and analyzes track aligned extents (traxtents ) extents that are aligned and ....

....use regular access patterns and large files. Although sequential full file access is relatively common [1, 29, 45] most data objects are much smaller than the disk request sizes needed to achieve good disk efficiency. For example, most files are well below 32 KB in size in UNIX like sys tems [15, 40] and below 64 KB in Microsoft Windows systems [12, 45] Directories and file attribute structures are almost always much smaller. To achieve sufficiently large disk requests in such environments, access patterns . a) System s view of storage. b) Mapping of LBNs onto physical ....

[Article contains additional citation context not shown here]

Gregory R. Ganger and M. Frans Kaashoek. Embedded inodes and explicit grouping: exploiting disk bandwidth for small files. Annual USENIX Technical Conference (Anaheim, CA), pages 1 17, January 1997.

No context found.

GANGER, G. R., AND KAASHOEK, M. F. Embedded inodes and explicit grouping: exploiting disk bandwidth for small files. In Proc. of the 1997.

No context found.

G. R. Ganger and M. F. Kaashoek. Embedded inodes and explicit groupings: Exploiting disk bandwidth for small files. In Proceedings of the 1997.

No context found.

G. R. Ganger and M. F. Kaashoek. Embedded inodes and explicit groupings: Exploiting disk bandwidth for small files. In Proceedings of the 1997.

No context found.

GANGER, G. R., AND KAASHOEK, M. F. Embedded inodes and explicit grouping: Exploiting disk bandwidth for small files. In

No context found.

G. R. Ganger and M. F. Kaashoek. Embedded inodes and explicit grouping: Exploiting disk bandwidth for small files. In Proceedings of the 1997.

No context found.

GANGER, G. R., AND KAASHOEK, M. F. Embedded inodes and explicit grouping: exploiting disk bandwidth for small files. In Proc. of the 1997.

No context found.

G. R. Ganger and M. F. Kaashoek. Embedded inodes and explicit groupings: Exploiting disk bandwidth for small files. In Proceedings of the 1997.

No context found.

G. R. Ganger and M. F. Kaashoek. Embedded inodes and explicit groupings: Exploiting disk bandwidth for small files. In Proceedings of the 1997.

No context found.

G. R. Ganger and M. F. Kaashoek. Embedded inodes and explicit grouping: Exploiting disk bandwidth for small files. In Proc. of the USENIX Annual Technical Conference, Anaheim, California USA, pages 1--17, January 1997.

No context found.

Gregory R. Ganger and M. Frans Kaashoek, "Embedded Inodes and Explicit Grouping: Exploiting Disk Bandwidth for Small Files," in Proceedings of the USENIX Annual Technical Conference, pages 1--17, 1997.

No context found.

G. R. Ganger and M. F. Kaashoek. Embedded inodes and explicit grouping: Exploiting disk bandwidth for small files. In Proceedings of the 1997.

No context found.

G. R. Ganger and M. F. Kaashoek. Embedded inodes and explicit grouping: Exploiting disk bandwidth for small files. In Proceedings of the 1997.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC