-
حرية الوصول المقاله
1 - تخصیص ایستای وظایف در سیستمهای توزیعشده با استفاده از الگوریتم ژنتیک موازی
Monire Taheri Sarvetaminدر طی دو دهه اخیر، بالارفتن فوقالعاده سرعت شبکههای رایانه ای و همچنین افزایش نیاز به سیستمهایی با کارایی بالا سبب شده است که محققان به پردازشهای موازی و توزیعشده علاقهمند شوند. رشد سریع سیستمهای توزیعشده باعث شده که مسائل گوناگونی در این زمینه مطرح شود. یکی از م أکثردر طی دو دهه اخیر، بالارفتن فوقالعاده سرعت شبکههای رایانه ای و همچنین افزایش نیاز به سیستمهایی با کارایی بالا سبب شده است که محققان به پردازشهای موازی و توزیعشده علاقهمند شوند. رشد سریع سیستمهای توزیعشده باعث شده که مسائل گوناگونی در این زمینه مطرح شود. یکی از مهمترین مسائلی که موردتوجه محققان زیادی قرارگرفته، مسئله تخصیص وظایف در اینگونه محیطها است که بهمنظور به دست آوردن بهرهوری مؤثر از سیستم انجام میشود. مسئله تخصیص وظایف بهجز در معدود موارد خاص جز مسائل NP-کامل است؛ بنابراین از فرایندهای اکتشافی برای دستیابی به راهحلهای زیربهینه در مدتزمان مطلوب استفاده میشود. اگرچه از روشهای مختلف در تحقیقات استفادهشده، اما هنوز پیدا کردن روش مؤثر و کارا برای این مشکل موردنیاز و مطلوب است. در این پژوهش از الگوریتم ژنتیک موازی برای پیدا کردن راهحل بهینه برای تخصیص یک گراف از وظایف به پردازندهها در سیستم توزیعشده استفادهشده است. نتایج نشان داد الگوریتم پیشنهادی میتواند تخصیصهای بهینه یا نزدیک بهینه برای مسائل با اندازههای گوناگون ارائه دهد. همچنین روش پیشنهادی توانست در زمان بسیار سریعتر از الگوریتم ژنتیک سنتی و با تسریع فراخطی، مسائل با اندازههای بزرگ و متوسط را حل کند. تفاصيل المقالة -
حرية الوصول المقاله
2 - ارائه روشی جدید برای شناسایی و از بین بردن پراسسهای یتیم
آرش صباغیدر سیستمهای توزیع شده به دلیل انجام فراخوانی پروسیجرهای راه دور، امکان ایجاد پراسسهای یتیم وجود دارد. در واقع پراسسهای یتیم، محاسبات راه دوری هستند که فراخواننده آنها به هر دلیل متوقف شده باشد. وجود محاسبات یتیم در سیستم به صورت کلی مطلوب نیست چراکه منجر به هدر رفت م أکثردر سیستمهای توزیع شده به دلیل انجام فراخوانی پروسیجرهای راه دور، امکان ایجاد پراسسهای یتیم وجود دارد. در واقع پراسسهای یتیم، محاسبات راه دوری هستند که فراخواننده آنها به هر دلیل متوقف شده باشد. وجود محاسبات یتیم در سیستم به صورت کلی مطلوب نیست چراکه منجر به هدر رفت منابع میشوند و همچنین این امکان وجود دارد که وضعیتهای ناسازگاری را در داده اشتراکی ببینند. در این مقاله روشی جدید به منظور شناسایی و از بین بردن پراسسهای یتیم ارائه شده است. این روش دارای مزایای قابل توجهی نسبت به روشهای مطرح موجود در این حوزه میباشد. از مهم ترین مزایای این روش میتوان به این موارد اشاره نمود: از بین بردن تمامی فرزندان و نوادگان راه دور پراسس یتیم ،تعداد اندک پیام ،عدم کاهش کارایی،توازن بار و افزایش قابلیت تحمل خطا. تفاصيل المقالة -
حرية الوصول المقاله
3 - An Improved Token-Based and Starvation Free Distributed Mutual Exclusion Algorithm
Om-Kolsoom Shahryari Ali BroumandniaDistributed mutual exclusion is a fundamental problem of distributed systems that coordinates the access to critical shared resources. It concerns with how the various distributed processes access to the shared resources in a mutually exclusive manner. This paper presen أکثرDistributed mutual exclusion is a fundamental problem of distributed systems that coordinates the access to critical shared resources. It concerns with how the various distributed processes access to the shared resources in a mutually exclusive manner. This paper presents fully distributed improved token based mutual exclusion algorithm for distributed system. In this algorithm, a process which has owing token, could enter to its critical section. The processes communicate to each other in an asynchronous message passing manner. We assume the distributed processes are organized in a wraparound two dimensional array. Also, the communication graph of the network is supposed to be a complete graph. The proposed algorithm uses three types of messages, namely ReqMsg, InfoMsg and RelMsg. Beside token-holding node, there are some nodes, we call them informed-nodes, which can know token-holding node and transmit request message to it directly. The number of messages, which are exchanged per each critical section entrance, is a key parameter to avoid posing additional overhead to the distributed system. In this paper, we obtain to messages per critical section access where N is the number of nodes in the system. The proposed algorithm outperforms other token based algorithms whilst fairness is kept and the proposed algorithm is starvation free. تفاصيل المقالة -
حرية الوصول المقاله
4 - Hybrid Meta-heuristic Algorithm for Task Assignment Problem
Mohammad Jafar Tarokh Mehdi Yazdani Mani Sharifi Mohammad Navid MokhtarianTask assignment problem (TAP) involves assigning a number of tasks to a number of processors in distributed computing systems and its objective is to minimize the sum of the total execution and communication costs, subject to all of the resource constraints. TAP is a أکثرTask assignment problem (TAP) involves assigning a number of tasks to a number of processors in distributed computing systems and its objective is to minimize the sum of the total execution and communication costs, subject to all of the resource constraints. TAP is a combinatorial optimization problem and NP-complete. This paper proposes a hybrid meta-heuristic algorithm for solving TAP in a heterogeneous distributed computing system. To compare our algorithm with previous ones, an extensive computational study on some benchmark problems was conducted. The results obtained from the computational study indicate that the proposed algorithm is a viable and effective approach for the TAP. تفاصيل المقالة -
حرية الوصول المقاله
5 - Minimizing job execution time in Data Grid by A fuzzy dynamic replication algorithm
مهسا بیگرضایی سید لیلی میر طاهری mirtaheriThe nature of Data Grids is dynamic. In these systems, data access patterns of users and network latency may change. The system needs to meet data availability reliability. Data replication is a well-known method for improving performance parameters such as data access أکثرThe nature of Data Grids is dynamic. In these systems, data access patterns of users and network latency may change. The system needs to meet data availability reliability. Data replication is a well-known method for improving performance parameters such as data access time, availability, load balancing, and reliability. Here, a novel dynamic algorithm is proposed that uses fuzzy inference systems to manage replication for increasing performance. The proposed algorithm uses a new comprehensive set of decision parameters and fuzzy logic in each phase to reduce the inefficiency caused by wrong decisions in different phases in a practical Grid. The algorithm uses two fuzzy interfere systems to select an appropriate place for new replication and a less valuable file for deleting when storage space is full. It places the new replica in a suitable site where the file could possibly be needed soon with high probability. It also prevents deleting valuable files using a fuzzy valuation function. The algorithm was simulated by the OptorSim simulator. The results demonstrated that the algorithm is more effective than other replication methods in terms of the number of created replications, the percentage of storage used, and the job execution time. تفاصيل المقالة -
حرية الوصول المقاله
6 - Static Task Allocation in Distributed Systems Using Parallel Genetic Algorithm
Monire Taheri SarvtaminOver the past two decades, PC speeds have increased from a few instructions per second to several million instructions per second. The tremendous speed of today's networks as well as the increasing need for high-performance systems has made researchers interested in par أکثرOver the past two decades, PC speeds have increased from a few instructions per second to several million instructions per second. The tremendous speed of today's networks as well as the increasing need for high-performance systems has made researchers interested in parallel and distributed computing. The rapid growth of distributed systems has led to a variety of problems. Task allocation is a key process for distributed systems to achieve effective system efficiency, which, except for a few cases, is an NP-complete problem. Finding an effective and efficient method for this problem is still sought despite various methods used in studies. Experiments and the results of previous research have shown that NP problems are better solved by exploratory methods than other methods. This study used a parallel genetic algorithm (PGA) to find a solution for proper task allocation to processors in a distributed system. The task allocation policy, obtained using PGA, is much faster than traditional genetic algorithms. The results showed that the proposed algorithm can provide optimal or near-optimal allocations for problems of different sizes. The proposed method also solved large- and medium-sized problems much faster than traditional genetic algorithms and with super linear speedup. تفاصيل المقالة -
حرية الوصول المقاله
7 - Reliability Analysis of Distributed System for Enhancing Data Replication using Gumbel-Hougaard Family Copula Approach Joint Probability Distribution
Ibrahim Yusuf Abdulkareem Ismail Nasir Sufi Faruku Ambursa Abdullahi Sanusi Muhammad IsaIn today’s world of technology, it is impossible to see where computer system does not play an important role. The application of distributed systems is gradually becoming broad and diverse, and as a result of this, reliability prediction is a key concern. This pa أکثرIn today’s world of technology, it is impossible to see where computer system does not play an important role. The application of distributed systems is gradually becoming broad and diverse, and as a result of this, reliability prediction is a key concern. This paper, considered a distributed system with five standby subsystems A (the clients), B (two load balancers), C (two distributed database servers), D (two mirrored distributed database serves) and E (centralized database server) is considered arranged as series-parallel system. Exponential failure and repair are susceptible for all the components of this system. Each component’s failure rates are constant and considered to obey an exponential distribution, and they are repaired using general repair or copula repair. The system is evaluated using first-order partial differential equations and the supplementary variable technique, Gumbel-Hougaard family of Copula, to find expressions for reliability metrics of system strength such as availability, reliability, MTTF, sensitivity, and profit function. These reliability metrics have been validated for different parametric values and the results are presented in tables and figures. تفاصيل المقالة