A cooperative agents-based workflow-level distributed data placement strategy for scientific cloud workflows

Within the Cloud Computing context, the placement of massive data used by scientific workflows environment appears to be costly in terms of energy consumption and data transfer time. Indeed, due to the large size of consumed and generated datasets of scientific workflows tasks, the data placement problem becomes more challenging task. This problem is considered as NP-Hard problem. Ensuring an optimal mapping of data to Cloud Storage Services at a reasonable time turns out to be necessary. Many task-level approaches have been hence proposed while considering shared datasets within individual workflows to reduce data transfer cost, which is not efficient enough for the situation of multiple workflows. In the same perspective, taking into account the fixed datasets still an important issue. In the present paper, cooperative agents and Formal Analysis Concepts (FCA) based-data placement strategy for workflow-level data-intensive workflows is proposed. The proposed approach deals with three main concerns: i) treating multiple workflows simultaneously, ii) considering all types of data, specifically, fixed datasets, and iii) reducing the execution time of the data placement algorithm based on a set of cooperative intelligent agents. Experimental results show that the distribution of the proposed data placement strategy among a set of cooperative agents and the FCA approach can be more cost-effective than its task-level counterpart. Eventually, the proposed strategy proves to reduce the execution time during the execution of the tasks of multiple workflows.