Data can be replicated to improve availability and performance in distributed systems. To ensure one copy serializability, many protocols have been proposed. Most of these protocols are based on the quorum set approac...
详细信息
Data can be replicated to improve availability and performance in distributed systems. To ensure one copy serializability, many protocols have been proposed. Most of these protocols are based on the quorum set approach. However, some of them are dominated. Hence, these protocols can be further improved. We devise a paradigm for refining protocols for managing replicated data based on theoretical analysis. Examples are presented to illustrate the usefulness of our methodology.< >
MCM technology can be usefully exploited in a number of areas in the design of large scale distributed memory MIMD processors. Within the interconnection technologies for multichip assemblies (ITMA) project two demons...
详细信息
ISBN:
(纸本)0780308670
MCM technology can be usefully exploited in a number of areas in the design of large scale distributed memory MIMD processors. Within the interconnection technologies for multichip assemblies (ITMA) project two demonstrators have been selected for different applications within a parallel computer. the first demonstrator circuit is an extended switch element using a multichip module technology. the MCM usess four Elite switch devices connected on a silicon substrate. Flip chip bonding is used. A summary of device characteristics is given. the second demonstrator integrates a RISC CPU chip set along with a communications coprocessor. the resulting module contains all the components required for a MIMD processing node, withthe exception of the DRAM and DRAM controller. the two demonstrator circuits are discussed.
the workstation model of parallelprocessing presents specific challenges caused by the latency of the communications network and the workload imbalance that arises from the heterogeneity of the nodes. Data partitioni...
详细信息
the workstation model of parallelprocessing presents specific challenges caused by the latency of the communications network and the workload imbalance that arises from the heterogeneity of the nodes. Data partitioning is critically important for parallelprocessing in this environment. We mathematically characterize the communication costs for four data decomposition schemes: scatter, contiguous point, contiguous row, and block. these methods are analyzed in terms of problem size, number of processors, network speed, and communication pattern. Bounds are established for the performance of these decomposition schemes that can be used to make better-informed data partitioning decisions.< >
We propose a mechanism for improving the performance of name resolution operations in distributed file systems. the mechanism is based on the idea of reducing the number of message exchanges required for the resolutio...
详细信息
We propose a mechanism for improving the performance of name resolution operations in distributed file systems. the mechanism is based on the idea of reducing the number of message exchanges required for the resolution of a particular file name. In this mechanism, users are given the flexibility to dynamically define and change their performance requirements for the various file names being used by them.< >
In this paper, we propose the technique of lazy checkpoint coordination which preserves process autonomy while employing communication-induced checkpoint coordination for bounding rollback propagation. the notion of l...
详细信息
ISBN:
(纸本)0818643129
In this paper, we propose the technique of lazy checkpoint coordination which preserves process autonomy while employing communication-induced checkpoint coordination for bounding rollback propagation. the notion of laziness is introduced to control the coordination frequency and allow a flexible trade-off between the cost of checkpoint coordination and the average rollback distance. Worst-case overhead analysis provides a means for estimating the extra checkpoint overhead. Communication trace-driven simulation for several parallel programs is used to evaluate the benefits of the proposed scheme.
We introduce a memory storage scheme allowing conflict-free parallel access to rows, columns, square blocks, distributed blocks, and positive and negative diagonals of two dimensional arrays. Unlike the existing schem...
详细信息
We introduce a memory storage scheme allowing conflict-free parallel access to rows, columns, square blocks, distributed blocks, and positive and negative diagonals of two dimensional arrays. Unlike the existing schemes, the proposed scheme can be used for an arbitrary number of memory modules and an arbitrary size of the arrays. We develop a systematic procedure for the memory allocation based on a placement matrix constructed using circulant matrices.< >
We consider the following problem. Given a point z in the interior of a simple polygon P with n vertices, find all the boundary points of P that are "visible" from z. We present two parallel algorithms for t...
详细信息
We consider the following problem. Given a point z in the interior of a simple polygon P with n vertices, find all the boundary points of P that are "visible" from z. We present two parallel algorithms for this problem on a two dimensional reconfigurable mesh. the first algorithm runs in O(log/sup 2/ n) timing using n/sup 1/2/ processors, and the second one runs in O(1) time using n /spl times/ n processors.< >
Previously proposed load sharing algorithms do not support flexible sharing policies in a non-intrusive fashion and do not scale to systems consisting of several thousand workstations, and, therefore, are not amenable...
详细信息
Previously proposed load sharing algorithms do not support flexible sharing policies in a non-intrusive fashion and do not scale to systems consisting of several thousand workstations, and, therefore, are not amenable for owner-based distributed systems. the paper introduces a new algorithm that supports a rich set of policies while scaling to adequate system sizes with bounded intrusiveness.< >
this paper proposes and evaluates an approach for partitioning and mapping a class of parallel discrete event simulation models. the class of models is a subset of Petri Nets known as marked graphs. Six commercial sim...
详细信息
this paper proposes and evaluates an approach for partitioning and mapping a class of parallel discrete event simulation models. the class of models is a subset of Petri Nets known as marked graphs. Six commercial simulation models based on the marked graph modeling paradigm are examined to evaluate the available parallelism typically found in real models. the parallelism that can be exploited is very dependent upon effective partitioning and mapping algorithms to ensure a useful distribution of concurrently executable events. A partitioning and mapping algorithm is proposed in this paper, and its performance is evaluated.< >
In object oriented databases, objects are connected to each other by various kinds of relationships and form a general directed graph. the idea of leveled clustering is utilized to take advantage of the relationships ...
详细信息
In object oriented databases, objects are connected to each other by various kinds of relationships and form a general directed graph. the idea of leveled clustering is utilized to take advantage of the relationships and solve the problem of object allocation and storage in a distributed object oriented database. Both nonreplicative and replicative algorithms are developed to allocate objects to different sites and organize them in the storage of the local databases.< >
暂无评论