J. Huber, C. L. Elford, D. A. Reed, A. A. Chien, and D. S. Blumenthal. Ppfs: A high performance portable parallel file system. In Proc. 9th ACM International Conference on Supercomputing, pages 485--394, 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....accesses will benefit from local disks even the user does not provide any new access pattern information (Figure 7 (right) 6 Related Work The related work can be divided into five groups. One is parallel file systems, including IBM Vesta [11] and PIOFS [12] Intel Paragon [28] PPFS [21] and so on. These parallel file systems, either commercial or experimental, take advantage of parallel I O techniques, caching, prefetching etc to achieve significant performance improvement. The storage of these systems usually includes only 8 Figure 7. I O Time for data replication(left) and ....

J. Huber, C. Elford, D. Reed, A. Chien, and D. Blumenthal. Ppfs: A high performance portable parallel file system. In Proc. of the 9th ACM International Conference on Supercomputing, pages 385--394, 1995.

....easy for the user to change parameters directly in the Java window to get other prediction results. 6. Related work The related work can be divided into four groups. One is parallel file systems, including IBM Vesta [10] and PIOFS [11] Intel Paragon [24] HP Exemplar [5] Galley [23] PPFS [19], PIOUS [22] and so on. These parallel file systems, either commercial or experimental, take advantage of parallel I O techniques, caching, prefetching, etc. to achieve significant performance improvements. The storage of these systems usually includes only secondary storage resources and they are ....

J. Huber, C. Elford, D. Reed, A. Chien and D. Blumenthal, PPFS: A high performance portable parallel file system, in: Proc. of the 9th ACM International Conference on Supercomputing (1995) pp. 385-- 394.

....in a natural manner and still achieve good performance even on out of core problem sizes. 5 Related Work Much related work has depended on the use of an explicit I O interface by the programmer. On the OS side, this work includes the automatic detection of file access patterns in the file system [1, 10, 11, 12, 14, 15, 17], as well as the use of access patterns supplied directly by the application using an I O type of interface [22, 26, 30, 4] For compilers it involves analysis to move explicit I O calls back and change them to asynchronous I O calls instead [25] While some of the OS policies developed may be ....

J. Huber, C. L. Elford, D. A. Reed, A. A. Chien, and D. S. Blumenthal. PPFS: A high performance portable parallel file system. In Proceedings of the 9th A CM International Conference on Supercomputing, pages 385 394, Barcelona, July 1995.

....layer, whereas ERAID is designed to allow more functionality to be placed within the file system. Earlier work at HP on DataMesh also proposes more sophisticated interfaces for network attached storage [50] Our informed approach is also similar to a large body of work in parallel file systems [17, 24]. Most parallel file systems expose disk parallelism, but they allow the application itself, and not the file system, to manage it. Better control over redundancy in a parallel file system has also been proposed [9] In that work, the computation of parity is put under user control, and in doing ....

J. Huber, C. L. Elford, D. A. Reed, A. A. Chien, and D. S. Blumenthal. PPFS: A High Performance Portable Parallel File System. In Proceedings of the 9th ACM International Conference on Supercomputing, pages 385--394, Barcelona, Spain, July 1995.

....of I O access patterns in parallel systems. In [31] they improve on this approach by using hidden Markov models (HMMs) to perform classification, which provides more thorough control over policies than the neural network. This system is added to the Portable Parallel File System (PPFS) [19] in order to test the performance benefits in a real system. In all of these works they concentrate on the patterns of file access and how the caching and prefetching policies might be tuned to optimize. This is used throughout their I O system, at the server and client levels. In [28] they ....

....for the target platform. Development and implementation of such a system takes more time as the I O process code must be developed in addition to the user interfaces. The Portable Parallel File System (PPFS) is a user level file system designed for rapid experimentation on a variety of platforms[19, 50]. It implements various caching, prefetching, data placement striping, and data sharing policies. So called caching agents are used 34 to provide shared caches between application tasks and data servers which handle I O to underlying UNIX file systems. MPI or NXLIB is used to pass messages ....

Jay Huber, Christopher L. Elford, Daniel A. Reed, Andrew A. Chien, and David S. Blumenthal. PPFS: A high performance portable parallel file system. In Proceed- 139 ings of the 9th ACM International Conference on Supercomputing, pages 385--394, Barcelona, July 1995. ACM Press.

....Data transfer in recent cluster systems are achieved by extension of ftp. bbftp[5] sfcp[6] and GridFTP[7] are popular examples among scientific researchers. They can exploit communication speed by utilizing multi stream, parallel file transfer. But they require special API of middleware [18] or explicit use of multiple files in user programs. They also suffer overheads and bandwidth limitation mentioned above. The proposed Data Reservoir system utilizes performance from almost sequential disk accesses and efficiency of low level protocol during data transfer between distant nodes. ....

J. Huber, C.L.Elford, D.A.Reed, A.A. Chien and D.S.Blumenthal, "PPFS: A high performance portable parallel file system," Proc. of the 9th ACM Int. Conf. on Supercomputing, pp. 385-394, July 1995.

....that shows the best I O performance. 32 64 32 64 32 64 0.0 20.0 40.0 60.0 80.0 100.0 I O Bandwidth (MB Sec. Original) Level 1) Level 2 3) SDM Figure 7. I O bandwidth for RT 5. Related Work Several efforts have sought to optimize I O in parallel file systems and runtime libraries [3, 5, 6, 14, 16, 18, 22, 27, 31]. SRB (Storage Resource Broker) 2] provides an uniform interface to access various storage systems, such as file systems, Unitree, HPSS and database objects. However, it does not fully support the optimizations implemented in MPIIO. Shoshani et al. 28, 29] describe an architecture for op6 ....

J. Huber, C. L. Elford, D. A. Reed, A. A. Chien, and D. S. Blumenthal. PPFS: A High Performance Portable Parallel File System. In Proceedings of the 9th ACM International Conference on Supercomputing, pages 385--394. ACM Press, July 1995.

....in a natural manner and still achieve good performance even on out of core problem sizes. 5 Related Work Much related work has depended on the use of an explicit I O interface by the programmer. On the OS side, this work includes the automatic detection of file access patterns in the file system [1, 10, 11, 12, 14, 15, 17], as well as the use of access patterns supplied directly by the application using an I O type of interface [22, 26, 30, 4] For compilers it involves analysis to move explicit I O calls back and change them to asynchronous I O calls instead [25] While some of the OS policies developed may be ....

J. Huber, C. L. Elford, D. A. Reed, A. A. Chien, and D. S. Blumenthal. PPFS: A high performance portable parallel file system. In Proceedings of the 9th ACM International Conference on Supercomputing, pages 385--394, Barcelona, July 1995.

....the natural shared memory programming model and perform a large amount of inputs and outputs. To cope with this twofold requirement, a PSLS integrates a Shared Virtual Memory (SVM) 1, 12] thus providing programmers with a simple and attractive programming model, and a Parallel File System (PFS) [7, 16] to enable a large and efficient disk storage. PFSs usually provide high disk bandwith by fragmenting a file on several disks of different cluster nodes, enabling parallel accesses to fragments. Unfortunately, interfaces are quite complex which is contradictory with the presence of a simple shared ....

J.V. Huber, C.L. Elford, D.A. Reed, and A.A. Chien. PPFS: A high performance portable parallel file system. In Conference proceedings of the 1995 International Conference on Supercomputing, pages 385--394. ACM Press, July 1995.

....suggests that while parallel file systems can be useful, they do not avoid the need for careful parameter selection. Many specialized parallel I O libraries have been developed with the goals of providing portability, ease of use and high performance support to parallel applications. PPFS [7] focuses on efficient caching and prefetching support for parallel applications. MPI IO [3] provides a portable I O interface to MPI programs; it also supports collective I O interfaces. Parallel I O techniques such as two phase I O [2] and disk directed I O [9] for collective I O operations have ....

James V. Huber, Jr, Christopher L. Elford, Daniel A. Reed, Andrew A. Chien, and David S. Blumenthal. PPFS: A high performance portable parallel file system. In Proceedings of the International Conference on Supercomputing, June 1995.

....access and system parameters on the performance of the striped requests, we need to derive analytic models of disk striping. These models will also allow prediction of the input output behavior of peta scale machines with hundreds of thousands of disks. Several studies of parallel le systems [7, 8] have shown the importance of matching underlying le system policies with the application s access patterns. Mismatched policies and access patterns can signi cantly reduce input output performance. These observations point to several unresolved problems in dynamic le distributions ....

....selection of le striping distributions across storage devices and redundant storage of multiple distributions to reduce access time. This exploration builds atop our earlier work on physical and logical, input output pattern comparisons [9, 10] and portable, parallel le systems [7]. It integrates real time performance data, automatic access pattern classi cation, and fuzzy logic controls for choosing and con guring exible policies. The foundation of the research is a prototype software library called PPFS II (Portable Parallel File System II) In this paper, we outline an ....

[Article contains additional citation context not shown here]

J. V. Huber, C. L. Elford, D. A. Reed, A. A. Chien, and D. S. Blumenthal, \PPFS: A High-Performance Portable Parallel File System," in Proceedings of the 9th ACM International Conference on Supercomputing, pp. 385-394, July 1995.

....slower, because they are now using half the number of disks. The best choice for the number of disks to serve individual requests depends on a host of interrelated factors, including the access pattern, the system load, and hardware characteristics. Several studies on parallel file systems [10, 11] have shown the importance of matching underlying file system policies with the application s access patterns. Mismatched policies and access patterns can significantly reduce input output performance. Such a complex task requires file systems that can intelligently make adaptive file control ....

....load, To address these issues, this thesis investigates the performance directed selection of file striping distributions across storage devices. This exploration builds atop our earlier work on physical and logical input output pattern comparisons [12, 13] and portable, parallel file systems [10]. It integrates real time performance data, automatic access pattern classification, and fuzzy logic con4 trols for choosing and configuring flexible policies. The foundation of the research is a prototype software library called PPFS II (Portable Parallel File System II) 1.3 Thesis ....

[Article contains additional citation context not shown here]

J. V. Huber, C. L. Elford, D. A. Reed, A. A. Chien, and D. S. Blumenthal, "PPFS: A HighPerformance Portable Parallel File System," in Proceedings of the 9th ACM International Conference on Supercomputing, pp. 385--394, July 1995.

....is not guaranteed if multiple processes write to a common file [28] Client side caching must be disabled by locking the portion of the file being accessed, by using fcntl. A lock and unlock are therefore needed across the read write call. 4. Many research file systems provide their own APIs [9, 3, 11, 15, 20]. Implementing MPI IO on top of Unix I O functions will not be portable to these file systems. An alternative is to implement MPI IO on top of the POSIX I O interface [12] instead of the basic Unix I O functions. The POSIX interface is an international standard with greater functionality than ....

J. Huber, C. Elford, D. Reed, A. Chien, and D. Blumenthal. PPFS: A High Performance Portable Parallel File System. In Proceedings of the 9th ACM International Conference on Supercomputing, pages 385--394. ACM Press, July 1995.

....a factor of 2, and the speedup can be as high as a factor of 16. 6 Related Work Considerable effort has been put into building parallel I O libraries to provide high performance support for large scale scientific applications, such as Panda [27] DRA [24] PASSION [29] Jovian [4] and PPFS [17]. However, not much work has been done in providing automatic support for performance tuning in parallel I O systems. A recent effort has focused on automatically selecting efficient file system caching and prefetching policies in PPFS using two I O access pattern classification approaches, i.e. ....

J. Huber, C.L. Elford, D.A. Reed, A.A. Chien, and D.S. Blumenthal. PPFS: A high performance portable parallel file system. In Proceedings of the 9th ACM International Conference on Supercomputing, pages 385--394, Barcelona, July 1995. ACM Press.

....and atop the native Intel parallel file system (PFS) 4.1 Benchmarks Before experimenting with PPFS on real programs, it is useful to analyze it in the context of simple benchmarks. Using programs that consist solely of input output operations, we can compare PPFS with the native file system [13]. The benchmarks consist of two programs: one creates and writes a 64 MB file, and the other reads the same file. In the first case, each node writes a disjoint set of records. Writer i begins writing record i and then writes every Nth record, where N is the number of writers For the read test, ....

Huber, Jr., J. V., Elford, C. L., Reed, D. A., Chien, A. A., and Blumenthal, D. S. PPFS: A High Performance Portable Parallel File System. In 9th ACM International Conference on Supercomputing (submitted for publication 1994).

No context found.

J. Huber, C. L. Elford, D. A. Reed, A. A. Chien, and D. S. Blumenthal. Ppfs: A high performance portable parallel file system. In Proc. 9th ACM International Conference on Supercomputing, pages 485--394, 1995.

No context found.

J. Huber, C. L. Elford, D. A. Reed, A. A. Chien, and D. S. Blumenthal. Ppfs: A high performance portable parallel file system. In Proc. 9th ACM International Conference on Supercomputing, pages 485--394, 1995.

No context found.

J. Huber, C. Elford, D. Reed, and A. Chien. PPFS: A high performance portable parallel file system. In Conference proceedings of the 1995.

No context found.

J. Huber, C. Elford, D. Reed, and A. Chien. PPFS: A high performance portable parallel file system. In Conference proceedings of the 1995.

No context found.

Jay Huber, Christopher L. Elford, Daniel A. Reed, Andrew A. Chien, and David S. Blumenthal. PPFS: A high performance portable parallel file system. In Proceedings of Supercomputing '95, pages 385--394, Barcelona, July 1995.

No context found.

J. Huber, C.L. Elford, D.A. Reed, A.A. Chien, and D.S. Blumenthal, PPFS: A High Performance Portable Parallel File System, Proc. Nineth ACM Int'l Conf. Supercomputing, pp. 385394, July 1995.

No context found.

J. Huber, C. Elford, D. Reed, A. Chien, and D. Blumenthal, "PPFS: A high performance portable parallel file system. " ACM Int. Conf. Supercomputing, 1995.

No context found.

Jay Huber, Christopher L. Elford, Daniel A. Reed, Andrew A. Chien, and David S. Blumenthal. PPFS: A high performance portable parallel file system. In Proceedings of the 9th ACM International Conference on Supercomputing, pages 385--394, Barcelona, July 1995. ACM Press.

No context found.

Jay Huber, Christopher L. Elford, Daniel A. Reed, Andrew A. Chien, and David S. Blumenthal. PPFS: A high performance portable parallel file system. In Proceedings of the 9th ACM Conference on Supercomputing, pages 385--394, Barcelona, July 1995.

No context found.

James V. Huber, Christopher L. Elford, Daniel A. Reed, Andrew A. Chien, and David S. Blumenthal. PPFS: A high performance portable parallel file system. In Proceedings of the 9th ACM International Conference on Supercomputing, pages 385--394, Barcelona, July 1995. ACM Press.

First 50 documents  Next 50

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