Rajesh Bordawekar, Juan Miguel del Rosario, and Alok Choudhary. Design and evaluation of primitives for parallel i/o. In Proceedings of Supercomputing '93, pages 452--461, Portland, Oregon, November 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....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 been adopted in many of these libraries. However, no published studies examine the major performance factors of these systems for a wide range of I O patterns, problem sizes, and execution environments. Little work has been done on ....

Rajesh Bordawekar, Juan Miguel del Rosario, and Alok Choudhary. Design and evaluation of primitives for parallel i/o. In Proceedings of Supercomputing '93, pages 452--461, Portland, Oregon, November 1993.

....requests, it is often the case that in the aggregate the whole array is being written to or read from the file. The application can use this knowledge to significantly improve its I O performance. The technique of collective I O has been developed to better utilize the parallel I O subsystem [10, 26, 27, 4, 17, 23, 5, 8]. In this approach, the processors exchange information about their individual I O requests to develop a picture of the aggregate I O request. Based on this global knowledge, I O requests are combined and submitted in their proper order, making a much more efficient use of the I O subsystem. ....

Bordawekar, R., del Rosario, J. and A. Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing '93, pages 452--461, Portland, OR, 1993. IEEE Computer Society Press.

....requests, it is often the case that in the aggregate the whole array is being written to or read from the file. The application can use this knowledge to significantly improve its I O performance. The technique of collective I O has been developed to better utilize the parallel I O subsystem [10, 26, 27, 4, 17, 23, 5, 8]. In this approach, the processors exchange information about their individual I O requests to develop a picture of the aggregate I O request. Based on this global knowledge, I O requests are combined and submitted in their proper order, making a much more efficient use of the I O subsystem. ....

Bordawekar, R., del Rosario, J. and A. Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing '93, pages 452--461, Portland, OR, 1993. IEEE Computer Society Press.

....after the reordering process. In the New Mexico Order, the data array is striped across the four computation nodes in units 51 . A[113 114] A[1 2] A[15 16] Disk 0 Disk 1 Disk 2 Disk 3 Disk 4 Disk 5 Disk 6 Disk 7 A[7 8] A[21 22] A[5 6] A[19 20] A[3 4] A[17 18] A[115 116] A[119 120] A[127 128] A[117 118] A[9 10] A[23 24] A[121 122] A[11 12] A[25 26] A[123 124] A[13 14] A[27 28] A[125 126] Parity Parity Parity Parity Figure 8.3: Data Layout on the Disk Array of 14 data elements (i.e. the blocking ....

....magnitude faster than the Delta. Intel is redressing this differential with the Parallel File System (PFS) on the Intel Paragon XP S [19] PFS supports many of the same features as SFS, including disk striping, parity for fault tolerance, and distributed file caching. Recently, Bordawekar et al. [6] proposed a two phase access strategy to improve input output performance on the Delta and other parallel file systems. The first phase reads data from the disks with the data distribution and stripe size that best matches the distribution of the data stored on disk; the second phase rearranges ....

Bordawekar, R., del Rosario, J. M., and Choudhary, A. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing 1993 (Nov 1993), pp. 452-- 461.

.... been a major bottleneck in these applications [18, 8] Collective I O, where all processes cooperate to carry out largescale I O operations, has been suggested as a way to get around the I O bottleneck in many parallel applications, and several runtime libraries offer collective I O facilities [20, 2, 17]. However, these libraries are originally designed for applications running on massively parallel platforms; they implicitly assume that all processors are homogeneous and thus I O servers are placed in a predetermined manner (e.g. the first k processors will have I O servers) with no ....

R. Bordawekar, J. Rosario, and A. Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing '93, pages 452--461, 1993.

....it off. 1 Introduction If you are already familiar with latex, we don t have anything new to tell you about latex; but you still might enjoy looking at the lists below. If you are new to latex, you might appreciate our examples of how to cite papers [Fox90] and of how to cite multiple papers [Bord93, Maier93, Galbreath93]. Or, you might like to examine some examples of lists. 1.1 Numbered Lists Here s a numbered list: 1. If you haven t explored the Internet yet, it s time to get started. 2. The best way to do it is to get your sysadmin to install Mosaic (ftp the appropriate binary from ftp.ncsa.uiuc.edu) the ....

Rajesh Bordawekar, Juan Miguel del Rosario, and Alok Choudary, `Design and Evaluation of Primitives for Parallel I/O,' Proceedings of Supercomputing '93, pages 452-461, 1993.

....that can improve performance significantly. These extensions allow users to perform bulk (array) I O operations with a single method call. We have implemented these extensions and validated their performance benefits. 2 1. 3 Related Work Other than the large body of work related to parallel I O [4, 8, 9, 13, 23, 27, 28, 32, 33], the work most closely related to ours is the Jaguar project [36, 37] which aims to improve Java I O performance as one of its goals. Jaguar allows the Java runtime system to be extended with new primitive operations that enable efficient access to hardware resources. These primitives are ....

Bordawekar, R., del Rosario, J., and Alok Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing '93, pages 452-461, Portland, OR, 1993. IEEE Computer Society Press.

....undoubtedly benefit from the proposed methods. Finally, we note that the proposed methods are not just useful for I O, but also for interprocess communication, and would therefore benefit networking applications as well. 7. RELATED WORK Other than the large body of work related to parallel I O [1, 4, 5, 8, 14, 16, 17, 20, 21], the work most closely related to ours is the Jaguar project [23, 24] which aims to improve Java I O performance as one of its goals. Jaguar allows the Java runtime system to be extended with new primitive operations that enable efficient access to hardware resources. These primitives are ....

Bordawekar, R., del Rosario, J., and Alok Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing '93, pages 452-461, Portland, OR, 1993. IEEE Computer Society Press.

....storage, retrieval and processing of very large multi dimensional datasets. An initial discussion of a framework for scientific data management similar to the one described in this paper is given in [6] Several efforts have involved optimizing I O in parallel file systems and runtime libraries [3, 4, 7, 13, 16, 18, 22, 27, 31]. However, file systems and libraries have a lower level interface than SDM, requiring more work from the user. 6 Conclusions and Future Work We have presented the design and implementation of an environment for high performance scientific data management, called Scientific Data Manager (SDM) ....

Rajesh Bordawekar, Juan Miguel del Rosario, and Alok Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing '93, pages 452--461, November 1993. 11

....at the IBM T. J. Watson Research Center [Corbett93] It provides a userdefinable view of parallel files that are applicable to array data. In our work, we utilize a higher level semantic interface than they provide, and expect our library will be layered above file systems like Vesta. [Bordawekar93] describes run time primitives to support a two phase access strategy for conducting parallel i o. Such a facility is useful, although it is not needed so far in the write intensive applications we studied. Galbreath93] reports on experiences with parallel applications at Argonne National ....

Rajesh Bordawekar, Juan Miguel del Rosario, and Alok Choudary, Design and Evaluation of Primitives for Parallel I/O, Proceedings of Supercomputing '93, pages 452-461, 1993.

....level (as done in this study) implementing them as part of the language should provide much better performance. It is also worth noting that such methods would also be very beneficial to networking applications. 7 Related Work Other than the large body of work related to parallel I O [1, 4, 5, 8, 14, 16, 17, 19, 20], the work most closely related to ours is the Jaguar project [22, 23] which has as one of its goals improvement in the performance of Java I O. Jaguar allows the Java runtime system to be extended with new primitive operations that enable efficient access to hardware resources. These primitives ....

Bordawekar, R., del Rosario, J. and Alok Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing '93, pages 452-461, Portland, OR, 1993. IEEE Computer Society Press.

....a high level semantic view of the collective i o requests, the file system is not able to predict whether sequential prefetching will be useful or when to flush the file cache. Intel CFS [Pierce93] uses traditional caching, and [Kotz93b] shows that CFS only uses half of the raw disk bandwidth. Bordawekar93] considers a two phase access strategy for collective i o. In this approach, for read operations, the compute nodes cooperate to bring all the data into memory in a way that minimizes the total number of disk accesses by having the data layout in memory conform to the data layout on disk. In ....

R. Bordawekar, J. Miguel del Rosario, and A. Choudary, Design and Evaluation of Primitives for Parallel I/O, Proceedings of Supercomputing '93, pages 452-461, 1993.

....column, these p p processes can each access n= p p data elements, while the other processes access 0 2 We assume that p is a square. 15 system advantages disadvantages nCUBE [16] simple partitioning based on bit permutations all sizes must be powers of 2 array partitioning library [7, 6, 23] supports common array partitioning patterns, high level of abstraction must access full array in one operation nested strided [50] supports the common multidimensional access patterns user needs to compute offsets and strides in all dimensions, must access a full multidimensional ....

....to store array data, the same partitioning scheme can be used. In effect, the distribution of the array data among the processes induces a partitioning of the file segment that stores the array. This has been suggested in a number of libraries, especially in the context of providing I O for HPF [7, 6, 23]. Naturally, it allows all the common partitioning patterns to be expressed. The interface supported by these libraries is a high level interface suitable for direct use by programmers, and using the same abstraction (i.e. partitioned arrays) An analogous low level interface has also been ....

R. Bordawekar, J. M. del Rosario, and A. Choudhary, "Design and evaluation of primitives for parallel I/O". In Supercomputing '93, pp. 452--461, Nov 1993.

....the system is able to obtain more semantic information about the operations and may improve their efficiency. This cannot be done if many small operations are requested at different instants of time. This basic idea has led some very interesting projects such as Two phase I O [HdC95, CBM 94, BdC93, dBC93] Jovian [BBS 94] and Disk directed I O [Kot95a, Kot95b, Kot94] Two phase I O This I O strategy involves a division of the parallel I O task into two separate phases. In the first phase, the parallel data access is performed using a data distribution, stripe size, and set of reading ....

Bordawekar, R. R., del Rosario, J. M., and Choudhary, A. N. Design and evaluation of primitives for parallel I/O. In Proceedings of the Supercomputing'93, pages 452--461. IEEE Computer Society Press,

....CM5 sfs [LOVE93] Paragon PFS [BERR94] Mach [ROY93] and Zebra [HART92] Although they are parallel distributed file systems, none of them places the special interest in caching we present in this paper. A different approach has been the design of specialized interfaces to support collective I O [BORD93] [CORB93] With a collective I O interface, all processes make a single joint request to the file system, rather than numerous independent requests. D. Kotz has taken the idea a little further introducing a file system support to the collective I O resulting in the Disk directed I O technique ....

R.R. Bordawekar, J.M. del Rosario and Alok N. Choudhary "Design and Evaluation of Primitives for Parallel I/O" Proceedings of Supercomputing'93, pp. 452-461

....Most are based on a fairly traditional Unix like interface, in which individual processes make a request to the file system for each piece of the file they read or write. Increasingly common, however, are specialized interfaces to support multidimensional matrices [CFPB93, SW94, GL91, GGL93, BdC93, BBS 94, Mas92] and interfaces that support collective I O [GGL93, BdC93, BBS 94, Mas92] With a collective I O interface, all processes make a single joint request to the file system, rather than numerous independent requests. Disk directed I O is a promising new technique that takes ....

....processes make a request to the file system for each piece of the file they read or write. Increasingly common, however, are specialized interfaces to support multidimensional matrices [CFPB93, SW94, GL91, GGL93, BdC93, BBS 94, Mas92] and interfaces that support collective I O [GGL93, BdC93, BBS 94, Mas92] With a collective I O interface, all processes make a single joint request to the file system, rather than numerous independent requests. Disk directed I O is a promising new technique that takes advantage of a collective I O interface, and leads to much better performance ....

Rajesh Bordawekar, Juan Miguel del Rosario, and Alok Choudhary. Design and evaluation of primitives for parallel I/O. In Proceedings of Supercomputing '93, pages 452--461, 1993.

....of magnitude faster than the Delta. Intel is redressing this differential with the Parallel File System (PFS) on the Intel Paragon XP S [7] PFS supports many of the same features as SFS, including disk striping, parity for fault tolerance, and distributed file caching. Recently, Bordawekar et al. [4] proposed a twophase access strategy to improve input output performance on the Intel Delta and other parallel file systems. The first phase reads data from the disks with the data distribution and stripe size that best matches the distribution of the data stored on disk; the second phase ....

Bordawekar, R., del Rosario, J. M., and Choudhary, A. Design and Evaluation of Primitives for Parallel I/O. In Supercomputing 1993 (Nov 1993), pp. 452--461.

....the ParServers of the file system. 3 Data Mapping in ParFiSys Usually the user vision of the file is a byte stream, whereas the file system vision is a set of scattered logical blocks. Moreover, in parallel and distributed file systems, the blocks may be spread out among several ION and devices [1]. Thus, a parallel file system as ParFiSys must be able to establish some correspondence between the user and the physical image of data. To satisfy user I O requests, each ParClient and ParServer must have some knowledge of where the data corresponding to the user image are located. Mapping ....

R. Bordawekar, J. Rosario, and A. Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Supercomputing 93, pages 452--462. IEEE, 1993.

....in parallel. The concatenation of the individual streams from each task forms the resulting stream for the entire array section. The mechanism adopted by DRMS of first redistributing the array and then having each task write local data is similar to the two phase access strategy described in [4]. We proceed to explain this operation in more detail. Figure 4: Example of the steps in array section streaming. Let x be a d dimensional slice, and let A be a d dimensional distributed array. Then A[x] represents a section of this distributed array. The first step in streaming A[x] consists ....

Bordawekar, R., del Rosario, J. M., and Choudhary, A. Design and evaluation of primitives for parallel I/O. In Proceedings of Supercomputing'93, Portland, OR, pages 452-461, November 1993.

....improving I O performance of 27 I O intensive (not just out of core) parallel applications. Chameleon was the first runtime system which provided extensive support for parallel I O [GGL93] del Rosario et al. proposed a twophase access strategy for efficient access of distributed arrays [dRBC93, BdRC93] This strategy was later extended by Kotz to optimize disk accesses [Kot94] PASSION runtime system builds on [BdRC93] and provides runtime routines to access distributed multidimensional arrays in the Local as well as Global Placement Models [CBD 95] Similar runtime projects include PANDA ....

.... system which provided extensive support for parallel I O [GGL93] del Rosario et al. proposed a twophase access strategy for efficient access of distributed arrays [dRBC93, BdRC93] This strategy was later extended by Kotz to optimize disk accesses [Kot94] PASSION runtime system builds on [BdRC93] and provides runtime routines to access distributed multidimensional arrays in the Local as well as Global Placement Models [CBD 95] Similar runtime projects include PANDA [SCJ 95] which uses array chunking to improve I O performance, PIOUS [MS94] which is a parallel I O runtime system ....

R. Bordawekar, J. del Rosario, and A. Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing'93, pages 452--461, November 1993. URL: ftp://erc.cat.syr.edu/ece/choudhary/PASSION/sc93.ps.Z.

....same distribution it is needed. For example, in an image processing application, images may arrive (even the network or from archives) in files. Each file will have to be redistributed into local files based on the out of core distribution. These routines use the two phase access strategy [dRBC93, BdRC93] which will read the data from the input data file using the most optimal access pattern (which depends on how the data is stored on disks) and redistribute the data over the processors using the high speed processor interconnection network. The processors will then store the data into ....

R. Bordawekar, J. del Rosario, and A. Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing'93, pages 452--461, November 1993.

....irregular data sets; that is, data sets that are accessed via indirections. The data is reorganized (sorted) on the fly to eliminate postprocessing time. In this method, the requesting processors cooperate in reading or writing data a process known as collective I O, first proposed in general [2, 3, 14]. Specifically, processors cooperate to combine several I O requests into fewer larger granularity requests, reorder requests so that the file is accessed in proper sequence, and eliminate duplicate I O requests for the same data. In addition, I O workload is partitioned among processors ....

....that it often generates fine grained data distribution requiring access to non contiguous locations in global array. Therefore, appropriate collective I O method is necessary to obtain high I O performance. The design of the collective I O library function is based on the two phase I O strategies [2, 14]. The basic idea behind two phase collective I O is at runtime to reorder the access patterns seen by the I O system such that the patterns are optimized. In other words, large number of small and disjoint I O requests are converted into small number and large contiguous requests. This ....

Rajesh Bordawekar, Juan Miguel del Rosario, and Alok Choudhary. Design and evaluation of primitives for parallel I/O. In Proceedings of Supercomputing '93, pages 452--461, 1993.

....characteristics to be considered to develop an I O library in order to support irregular problems. One of them is that irregular problems often generate fine grained data distributions requiring access to noncontiguous locations in a global array. Our library uses the two phase I O strategies [1] to convert a large number of small and disjoint I O requests into a small number of large contiguous requests. Several factors must be considered in the design of a library based on this technique: buffer size used by the library, communication schedule construction and reorganization, the number ....

Rajesh Bordawekar, Juan Miguel del Rosario, and Alok Choudhary. Design and evaluation of primitives for parallel I/O. In Proceedings of Supercomputing '93, pages 452--461, Portland, OR, 1993. IEEE Computer Society Press.

....from files often requires each processor to make a large number of small non contiguous data requests. This problem can be addressed by replacing small noncontiguous requests by large collective requests. This approach, known as Collective I O, has been found to work extremely well in practice [BdC93, Kot96, SCJ 95] In this paper, we describe implementation and evaluation of a collective I O prototype in a production parallel file system on the Intel Paragon. The prototype is implemented in the PFS subsystem of the Intel Paragon Operating System. We evaluate the collective I O ....

....patterns generated in parallel workloads [KN94] This often leads to situations where processors are forced to make a large number of small non contiguous data requests. It has been shown that in such cases, a technique called Collective I O can be effectively used to improve the I O performance [BdC93, Kot94] The primary focus of this work is to provide file system support for collective I O. In this paper, we describe the design and evaluation of a collective I O prototype in the Intel Paragon Parallel File System. The implementation is done in a production parallel file system by modifying ....

[Article contains additional citation context not shown here]

Rajesh Bordawekar, Juan Miguel del Rosario, and Alok Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing '93, pages 452--461, 1993.

....during program execution. In this paper, we propose a technique called the Extended Two Phase Method, for accessing sections of out of core arrays efficiently. This is an extension and generalization of the Two Phase Method for reading in core arrays from files, which was previously proposed in [7, 3]. The Extended Two Phase Method uses collective I O in which all processors cooperate to perform I O in an efficient manner by combining several I O requests into fewer larger requests, eliminating multiple disk accesses for the same data and reducing contention for disks. We describe the ....

....the I O performance may be very low because of lower granularity of accesses and multiple accesses for the same data. Therefore, it is necessary to use a more efficient technique for doing I O in out of core computations. In the case of in core arrays, Bordawekar, del Rosario and Choudhary [7, 3] have proposed a Two Phase Method for reading an entire array from a file in an efficient manner, into a distributed array in main memory. This method is found to give consistently good performance for all data distributions. In this paper, we propose an Extended Two Phase Method for accessing ....

[Article contains additional citation context not shown here]

R. Bordawekar, J. del Rosario, and A. Choudhary. Design and Evaluation of Primitives for Parallel I/O. In Proceedings of Supercomputing '93, pages 452--461, November 1993.

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