Craig S. Freedman, Josef Burger, and David J. DeWitt. SPIFFI-a scalable parallel file system for the Intel Paragon. IEEE Transactions on Parallel and Distributed Systems, 7(11):1185--1200, November 1996.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....proposed a construct, called a set iterator, that exploits asynchrony and nondeterminism in data accesses to reduce aggregate I O latency. The River project [3] uses efficient streaming from distributed hererogenous nodes to maximize I O performance for data intensive applications. Other research [8, 13, 19] exploited similar ideas at the file system level to achieve higher I O throughput from the underlying, inherently parallel, hardware. Our goal is to achieve maximum storage efficiency without changing the existing DBMS structure and policies. 2.3 Exploiting Observed Storage Characteristics To ....

Craig S. Freedman, Josef Burger, and David J. DeWitt. SPIFFI-a scalable parallel file system for the Intel Paragon. IEEE Transactions on Parallel and Distributed Systems, 7(11):11851200, November 1996.

....insures the users investments. We have investigated many different methods to construct our web server cluster, including parallel file systems and distributed file systems. Unfortunately, these technologies are not met al..l our requirements. Parallel file systems, such as SPIFFI for Intel Paragon[1] and Bridge[2] can provide parallel I O capability for file accesses, even to an individual file. However, the predefined striping groups restrict the flexibility of system configuration and reconfiguration. Distributed file systems, such as NFS[3] AFS[4] 5] and Coda[6] have been widely used in ....

Craig S. Freedman, Josef Burger and David J. DeWitt, SPIFFI---A Scalable Parallel File System for the Intel Paragon, In IEEE Transactions on Parallel and Distributed Systems, Vol. 7, No. 11, pp. 1185-1120, Nov. 1996

....on the distributed characteristic of files and directories, such as NFS[1] AFS[2]0 and Coda0. They should bear the network latency and indetermination, such as the disconnection operation in Coda. Parallel file systems focus on the performance of parallel I O, such as SPIFFI for Intel Paragon[5] and Bridge[6] They distribute a single file on different storage to improve the parallelism of I O. However, cluster file systems adopt the characteristics of parallel and distributed file system. They have high speed network connection, distributed data storage and file granulat parallel I O. ....

Craig S. Freedman, Josef Burger and David J. DeWitt, SPIFFI---A Scalable Parallel File System for the Intel Paragon, In IEEE Transactions on Parallel and Distributed Systems, Vol. 7, No. 11, pp. 1185-1120, Nov. 1996

....it maintain consistency of metadata and yield a block pointer when client request. Second, Zebra uses single cleaner to create empty segments. Last, Zebra stripes each log segment to all storage servers that limits maximum number of storage server that can yield efficient throughput. SPIFFI [12], a parallel file system, implemented on top of the local file system. It stripes file blocks over disk nodes in round robin scheme. SPIFFI allows programmer to control striping parameters when file is created. It does not update metadata immediately after data was modified; it will modify ....

Craig S. Freedman, Josef Burger, and David J. DeWitt, "SPIFFI-A Scalable Parallel File System for the Intel Paragon", IEEE Transaction on Parallel and Distributed Systems, Vol. 7, No. 11, November 1996.

....to any number of nodes, keeping the probability of data unavailability arbitrarily small. 1 Introduction Within a few years, scalability became the key word to modern storage systems, and to the next generation of applications and of computer systems in general, A al95] C al95] M97a c] [FBW97]. Fast processing of large collections of data, required by modern applications, crucially needs horizontal scalability, i.e. the ability to run over a fast network of share nothing sites, or nodes. Of special interest is horizontal scalability over commodity components, e.g. mass produced ....

....at network and switched multicomputers, KLR96] as well and some others, e.g. SPW90] and [JK93] among the earliest, targeting multiprocessor supercomputers. Other proposals seek vertical scalability, over 39 a fixed and rather small size multicomputer, e,g, 16 nodes, B al95] [FBW97], M97a] and [P97] Some of the implementations provide support for high availability through mirroring, B al] H96] Others propose a more or less delayed replication for disaster recovery, and lower level capabilities, e.g. for failover recovery [M96] M97] Fig. 7 Controlled reliability ....

Freedman, C., Burger, J., DeWitt, D. SPIFFI -- A Scalable Parallel File System for the Intel Paragon. Trans. on Par. and Distr. Syst.

....to each other for synchronization purpose. Additionally, they need not store their own programs, which results in a smaller design and a bigger amount of processes. SPIFFI (Scalable Parallel File System) SPIFFI is a high performance parallel file system that stripes files across multiple disks [16]. SPIFFI provides applications with a high level flexible interface including one individual and three shared file pointers. STARFISH STARFISH is a parallel file system simulator which ran on top of the Proteus parallel architecture simulator, which in turn ran on a DEC 5000 workstation. ....

Craig S. Freedman, Josef Burger, and David J. Dewitt. SPIFFI --- a scalable parallel file system for the Intel Paragon. IEEE Transactions on Parallel and Distributed Systems, 7(11):1185--1200, November 1996.

....our attention to the large body of work in parallel file systems. Most systems have focussed on extracting high performance from a set of uniform disks, including PPFS [65] Bridge [46] Panda [110] Galley [89] Vesta [36] Swift [28] CFS [90] SFS [84] the SIO specification [15] and SPIFFI [50]. Some common features include scatter gather transfers, asynchronous interfaces, layout control, prefetching, and caching support at the client or server or both. Most of these parallel file systems stripe data naively across the set disks in the I O subsystem, which can have undesirable ....

....of these parallel file systems stripe data naively across the set disks in the I O subsystem, which can have undesirable performance properties. Shared File Pointers: One interesting feature provided by some of these systems is the notion of a shared file pointer, as found in CFS [90] and SPIFFI [50]. With a shared file pointer, multiple processes on different machines can access a file concurrently in a consistent manner, as if sharing a local file pointer. Shared file pointers have some excellent performance properties. For example, when a group of processes is reading from a data ....

Craig S. Freedman, Josef Burger, and David J. Dewitt. SPIFFI --- a scalable parallel file system for the Intel Paragon. IEEE Transactions on Parallel and Distributed Systems, 7(11):1185-- 1200, November 1996.

....are not dealt with in our comparative analysis. Different performance of various schemes broaden the class of applications that might benefit from. Among these are modern database systems, more and more often needing large scalable databases, and for which the parallel access is already a must [FBW97], B al95] Many of these databases grow very rapidly. The already mentioned UPS database passed from 4 to 13 TB between 97 and 98, and many other similar examples are known. The multimedia servers also start using multicomputers and the success may make them scaling big [B al95] H96] In the ....

Freedman, C., Burger, J., DeWitt, D. SPIFFI -- A Scalable Parallel File System for the Intel Paragon. Trans. on Par. and Distr. Syst.

....integrity in case of a crash) and global shared files pointers. MS94] reports results for ethernet tests. PIOUS also does not focus on buffering and prefetching issues. In the realm of parallel processing, several parallel file systems with interesting ideas have emerged [Kot94, DSE88, DS89, FBD94, CBF93, TBC 94] These systems are often geared toward matrix operations, and a number of descriptors to encapsulate typical access patterns are described [Kot94, CBF93] SPIFFI [FBD94] implements global shared file pointers of varying flavors (multiple clients access the same file pointer) ....

.... parallel processing, several parallel file systems with interesting ideas have emerged [Kot94, DSE88, DS89, FBD94, CBF93, TBC 94] These systems are often geared toward matrix operations, and a number of descriptors to encapsulate typical access patterns are described [Kot94, CBF93] SPIFFI [FBD94] implements global shared file pointers of varying flavors (multiple clients access the same file pointer) SPIFFI also implements a form of prefetching to handle the out of order requests generated when multiple clients sequentially access the same file. Their approach seems to use larger disk ....

Craig S. Freedman, Josef Burger, and David J. Dewitt. SPIFFI --- a scalable parallel file system for the Intel Paragon. Submitted to IEEE TPDS, 1994.

....on today s multiprocessors. When they find a multiprocessor that is configured with sufficient parallel I O hardware (unfortunately, many are not) they often discover that the file system software is not designed to meet their needs [CK93, KN94, PEK 95] or has poor performance [Nit92, KR94, FBD94] As a result, there are several proposals for new interfaces, run time libraries, compilers, languages, and file systems to support parallel applications on parallel computers. The focus of this paper is on a file system technique called disk directed I O, which can dramatically improve the ....

Craig S. Freedman, Josef Burger, and David J. Dewitt. SPIFFI --- a scalable parallel file system for the Intel Paragon. Submitted to IEEE TPDS, 1994.

....environment. While our focus in this paper is the improvement of the interface to a linear file model, the enhancement or outright replacement of that model is worthy of further investigation. One common enhancement to the conventional interface is a shared file pointer [Pie89, RP95, BGST93, FBD94] which provides a mechanism for regulating access to a shared file by multiple processes in a single application. The simplest shared file pointer is one which supports an atomic append mode (as in [LMKQ89] page 174) Intel s CFS provides this mode in addition to several more structured access ....

Craig S. Freedman, Josef Burger, and David J. Dewitt. SPIFFI --- a scalable parallel file system for the Intel Paragon. Submitted to IEEE TPDS, 1994.

....processors (CP) and I O processors (IOP) concurrent small requests from many processors, instead of the single large request that would have occurred on a uniprocessor. Indeed, since most multiprocessor file systems [CF94, FPD93, Pie89, Roy93, DdR92, LIN 93, BGST93, Kri94, Dib90, DSE88, FBD94] decluster file data across many disks, each application request may be broken into even smaller requests that are sent to different IOPs. It is difficult for the file system, which is distributed across many I O processors, to recognize these requests as a single coordinated request, and to use ....

Craig S. Freedman, Josef Burger, and David J. Dewitt. SPIFFI --- a scalable parallel file system for the Intel Paragon. Submitted to IEEE TPDS, 1994.

....expected demand for the content. Tiger, by contrast, stripes all content, eliminating the need for additional replicas to satisfy changing load requirements. Berson94] proposes an independently developed single machine disk striping algorithm with some similarities to that used by Tiger. SPIFFI [Freedman96] is a parallel file system implemented on an Intel Paragon system that can stripe data across large numbers of disks and can be used for multimedia files (as well as for more traditional parallel filesystem tasks) There is a certain similarity between the coherent hallucination model and ....

C. S. Freedman, J. Burger and D. J. DeWitt. SPIFFI -- A Scalable Parallel File System for the Intel Paragon. In IEEE Trans. on Parallel and Distributed Systems, 7(11), pages 1185-1200, November 1996.

....to provide parallel access to files. PFS stripes files across the I O nodes, the default stripe factor being 64 KB. We note that while PFS provides reasonable performance, it would appear that alternative file system organizations could provide better performance in some situations (e.g. see [11]) The Argonne IBM SP has 128 Power 1 processors connected by a multistage crossbar; 8 of these have an attached disk drive and are used for both computation and I O. The SP runs the AIX operating system and uses IBM s Parallel I O File System (PIOFS) to provide parallel access to files striped ....

....easily in other directions. For example, the matrices constructed by computational chemistry codes are often quite sparse and hence may benefit from the use of compression [7] Compression is easily incorporated into the DRA framework, as are other optimizations proposed for collective I O [31,8,11]. Acknowledgments Work at PNNL was performed under the auspices of the High Performance Computing and Communications Program of the Office of Computational and Technology Research, U.S. Department of Energy under contract DEAC066 76RLO1830 with Battelle Memorial Institute which operates the ....

C. Freedman, J. Burger, and D. DeWitt. SPIFFI -- a scalable parallel file system for the Intel Paragon. IEEE Trans. Parallel and Dist. Systems, 1996. to appear.

....and on the underlying technique of disk directed I O. 2 Background File systems. There are many parallel file systems today, including Bridge [DSE88] Intel CFS [Pie89] Intel PFS [Roy93] IBM Vesta [CF94] nCUBE [DdR92] TMC sfs [LIN 93, BGST93] Hurricane File System [Kri94] and SPIFFI [FBD95] There are also several systems intended for workload clusters, such as PIOUS [MS94] and VIP FS [dHC94] All of these systems decluster file data across many disks to provide parallel access to the data of any file. Workload. The CHARISMA project traced production parallel scientific computing ....

Craig S. Freedman, Josef Burger, and David J. Dewitt. SPIFFI --- a scalable parallel file system for the Intel Paragon. Submitted to IEEE TPDS.

No context found.

Craig S. Freedman, Josef Burger, and David J. DeWitt. SPIFFI-a scalable parallel file system for the Intel Paragon. IEEE Transactions on Parallel and Distributed Systems, 7(11):1185--1200, November 1996.

No context found.

Craig S. Freedman, Josef Burger and David J. DeWitt, SPIFFI---A Scalable Parallel File System for the Intel Paragon, In IEEE Transactions on Parallel and Distributed Systems, Vol. 7, No. 11, pp. 1185-1120, Nov. 1996

No context found.

C. S. Freedman, J. Burger, D. J. DeWitt. "SPIFFI -- a scalable parallel file system for the Intel Paragon," IEEE Trans. on Parallel and Distributed Systems, 7(11), pp. 1185-1200, Nov 1996.

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