Scholarlink

<p>Centralized workflow execution engines exhibit several common weaknesses, including bottlenecks, single points of failure, poor performance, unreliability, and limited scalability. Decentralized workflow execution engines have been introduced to address these issues. Moreover, cloud computing has been embraced to accommodate the growing requests and the escalating demand for additional resources. Consequently, the provision of distributed workflow engines as a service in cloud computing can effectively meet these requirements. Despite the continuous changes occurring in cloud computing runtime, workflows must remain adaptable to environmental fluctuations and be continuously configured based on the dynamics of the runtime environment. Consequently, researching the adaptability of decentralized workflow engines in cloud computing is paramount. Dynamic and adaptable fragmentation of workflows represents one of the methods to enhance the adaptability of the workflow management system. This research delves into two aspects of runtime workflow fragmentation concerning adaptability with the runtime of cloud computing: First, the adaptability of the number of created fragments to the number of virtual machines (referred to as fragment-proportionality). Secondly, the adaptability of the number of generated fragments is based on the current conditions of the communicative media (referred to as available-bandwidth). A fuzzy algorithm has also been proposed to select appropriate fragments, considering both adaptability aspects. An analysis of test results from a reference workflow shows that our method significantly boosts throughput, response time, and message exchange volumes compared to fully decentralized configurations. Each adaptability aspect individually enhances baseline performance. The Fuzzy algorithm applies both adaptability aspects. With variable bandwidth and constant virtual machine numbers, the algorithm resulted in response time and throughput improvements of [9.16–97.43%] and [4.9–306.53%]. It also led to response time and throughput enhancements of [27.27–84.26%] and [67.61–79.74%] with constant bandwidth with variable virtual machine numbers.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Adaptable decentralized workflow execution with fuzzy framework in cloud computing (ADWEF.Cloud)

  • 期刊论文

  • 发表日期：

    2025年5月22日

查看全文
  • Faramarz Safi-Esfahani,
  • Narges Khatibi

摘要

Centralized workflow execution engines exhibit several common weaknesses, including bottlenecks, single points of failure, poor performance, unreliability, and limited scalability. Decentralized workflow execution engines have been introduced to address these issues. Moreover, cloud computing has been embraced to accommodate the growing requests and the escalating demand for additional resources. Consequently, the provision of distributed workflow engines as a service in cloud computing can effectively meet these requirements. Despite the continuous changes occurring in cloud computing runtime, workflows must remain adaptable to environmental fluctuations and be continuously configured based on the dynamics of the runtime environment. Consequently, researching the adaptability of decentralized workflow engines in cloud computing is paramount. Dynamic and adaptable fragmentation of workflows represents one of the methods to enhance the adaptability of the workflow management system. This research delves into two aspects of runtime workflow fragmentation concerning adaptability with the runtime of cloud computing: First, the adaptability of the number of created fragments to the number of virtual machines (referred to as fragment-proportionality). Secondly, the adaptability of the number of generated fragments is based on the current conditions of the communicative media (referred to as available-bandwidth). A fuzzy algorithm has also been proposed to select appropriate fragments, considering both adaptability aspects. An analysis of test results from a reference workflow shows that our method significantly boosts throughput, response time, and message exchange volumes compared to fully decentralized configurations. Each adaptability aspect individually enhances baseline performance. The Fuzzy algorithm applies both adaptability aspects. With variable bandwidth and constant virtual machine numbers, the algorithm resulted in response time and throughput improvements of [9.16–97.43%] and [4.9–306.53%]. It also led to response time and throughput enhancements of [27.27–84.26%] and [67.61–79.74%] with constant bandwidth with variable virtual machine numbers.