A Holistic Approach for Collaborative Workload Execution in Volunteer Clouds

Abstract

e demand for provisioning, using and maintaining distributed computational resources is growing hand in hand with the quest for ubiquitous services. Centralized infrastructures such as cloud computing systems provide suitable solutions for many applications, but their scalability could be limited in some scenarios, e.g., in case of latency-dependent applications. e volunteer cloud paradigm aims at overcoming this limitation by encouraging clients to o er their own spare, perhaps unused, computational resources. Volunteer clouds are thus complex, large-scale, dynamic systems which demand for self-adaptive capabilities to o er e ective services, as well as modeling and analysis techniques to predict their behavior. In this paper, we propose a novel holistic approach for volunteer clouds supporting collaborative task execution services able to improve the QoS of compute-intensive workloads. We instantiate our approach by extending a recently proposed ant colony optimization algorithm for distributed task execution with a workload-based partitioning of the overlay network of the volunteer cloud. Finally, we evaluate our approach using simulation-based statistical analysis techniques on a workload benchmark provided by Google. Our results show that the proposed approach outperforms some traditional distributed task scheduling algorithms in the presence of compute-intensive workloads. e increasing role of so ware and the pervasiveness of computational devices have fostered the interest and demand for storage and computing services. Cloud computing is a recent but already well-established paradigm envisioning the use of computing as utility. anks to the cloud, users do not need to buy or maintain powerful computational devices, since they can use services remotely. Such services are customized according to the users' needs and requirements, and are provided transparently from where the resources are physically hosted. Another bene t of cloud systems is their ability to dynamically scale whenever application requirements change. In order to properly manage their resources, cloud providers o er contract-based services whose terms are speci ed in a Service Level Agreement (SLA), containing the required ality of Service (QoS) and the associated penalty fees in case of their violation.One of the greatest challenges in the cloud computing domain is the e cient use of resources, while guaranteeing the ful llment of SLAs. To this aim, particular a ention is given to autonomic computing techniques, which support the development of clouds as self-managing systems. Cloud features should automatically adapt to their internal status and the dynamically changing environment, without human intervention. Self-management may involve system maintenance, awareness, evolution, con guration, healing, tuning and optimization.Along with the latest advancements in virtualization technologies, the last decade has also witnessed a great leap forward in processing power, thanks to the advent of the multicore era. In turn, this scenario has brough...

Energy-Efficient Job-Assignment Policy With Asymptotically Guaranteed Performance Deviation

“…. , J), each of which has an arrival rate λ j > 0 indicating the average number of arrivals per unit time, following a Poisson process as previous studies in Cloud environments [29], [30]. If groups/type of Internet/network customers decide to send requests independently and identically during a given time period and the number of such customers is sufficiently large for the corresponding dynamic process to become stationary, then it is reasonable to model the arrival process of customer requests for this type as a Poisson process, although the arrival rates may vary from one time period to another, which is consistent with observations of real-world tracelogs [31], [32].…”

Graph Deep Factors for Forecasting

“…Hence, optimizing their use is beneficial for both cloud operators and tenants. Accurate forecasting of workload patterns in cloud can aid in resource allocation, scheduling, and workload co-location decision for different workloads (Sliwko, 2018;Sebastio et al, 2018;Sîrbu & Babaoglu, 2015), massively reducing the operating costs and increasing resource efficiency (Liu et al, 2018). Learning must be fast to ensure precise decisionmaking and quick adaption to changes in demand.…”

Guest Editorial for the Special Issue on FORmal methods for the quantitative Evaluation of Collective Adaptive SysTems (FORECAST)

“…One paper was rejected, while all others underwent two rounds of reviews. We have three papers that have undergone the Replicating Computational Results (RCR) review process [1,6,10]. Their respective RCR reports are included in this special issue [2,8,11].…”

“…The article by Sebastio et al [10] proposes a novel holistic approach to task scheduling in volunteer clouds. Volunteer clouds are decentralized cloud systems based on collaborative task execution, in which clients voluntarily share their spare computational resources to overcome the scalability limitations of cloud computing.…”