TY - GEN
T1 - Efficient Parallelism vs Reliable Distribution: A Trade-off for Concurrent Computations.
T2 - 5th International Conference on Concurrency Theory, CONCUR 1994
AU - Kanellakis, Paris C.
AU - Michailidis, Dimitrios
AU - Shvartsman, Alex A.
N1 - DBLP License: DBLP's bibliographic metadata records provided through http://dblp.org/ are distributed under a Creative Commons CC0 1.0 Universal Public Domain Dedication. Although the bibliographic metadata records are provided consistent with CC0 1.0 Dedication, the content described by the metadata records is not. Content may be subject to copyright, rights of privacy, rights of publicity and other restrictions.
PY - 1994
Y1 - 1994
N2 - Concurrent computations should combine efficiency with reliability, where efficiency is usually associated with parallel and reliability wiih distributed computing. Such a desirable combination is not always possible, because of an intuitive trade-o:ff: efficiency requires removing redundancy from computations whereas reliability requires some redundancy. We survey a spectrum of algorithmic models (from fail-stop, synchronous to asynchronous and from approximate to exact computations) in which reliability is guaranteed with small trade-o:ffs in efficiency. We illustrate a number of cases where optimal trade-o:ffs are achievable. A basic property of all these models, which is of some interest in the study of concurrency, is that “true” read/write concurrency is necessary for fault tolerance. In particular, we outline (from [14]) how algorithms can be designed so that, in each execution, the total “true” concurrency used can be closely related to the faults that can be tolerated.
AB - Concurrent computations should combine efficiency with reliability, where efficiency is usually associated with parallel and reliability wiih distributed computing. Such a desirable combination is not always possible, because of an intuitive trade-o:ff: efficiency requires removing redundancy from computations whereas reliability requires some redundancy. We survey a spectrum of algorithmic models (from fail-stop, synchronous to asynchronous and from approximate to exact computations) in which reliability is guaranteed with small trade-o:ffs in efficiency. We illustrate a number of cases where optimal trade-o:ffs are achievable. A basic property of all these models, which is of some interest in the study of concurrency, is that “true” read/write concurrency is necessary for fault tolerance. In particular, we outline (from [14]) how algorithms can be designed so that, in each execution, the total “true” concurrency used can be closely related to the faults that can be tolerated.
UR - http://www.scopus.com/inward/record.url?scp=37349108235&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=37349108235&partnerID=8YFLogxK
U2 - 10.1007/978-3-540-48654-1_21
DO - 10.1007/978-3-540-48654-1_21
M3 - Conference contribution
AN - SCOPUS:37349108235
SN - 9783540583295
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 242
EP - 266
BT - CONCUR 1994
A2 - Jonsson, Bengt
A2 - Parrow, Joachim
PB - Springer Verlag
Y2 - 22 August 1994 through 25 August 1994
ER -