This paper outlines a project, started in October 2023, and entitled Methodologies for the Parallelization, Performance Evaluation and Scheduling of Applications for the Cloud-Edge Continuum (MPESACC), aimed at developing a holistic approach to parallelize sequential code across the Cloud-Edge Continuum. MPESACC entails the formulation of a Directive-based programming model, integration of performance analysis, bottleneck identification, scheduling, and optimization techniques for the execution of parallelized code. The proposed model, tailored for Cloud-Edge computing, involves the design of a Compiler software capable of decomposing serial codes, annotated with directives, into distributed components for remote execution. Parallelization techniques conveyed through Parallel Patterns and communication templates in the form of Code Skeletons, simplify transformations and adhere to best practices in code parallelization and distribution. The performance analysis models consider crucial aspects of the Cloud-Edge continuum, such as bandwidth, processing capacity, and energy constraints, providing essential feedback for directive and pattern selection. Scheduling considerations address resource availability and computational needs, optimizing time, energy, storage, and bandwidth constraints. Robust mathematical models support scheduling optimization to prevent task failure or idleness due to temporary resource shortages. The proposed Directives guide scheduling algorithms by providing execution constraints, ensuring informed decisions for optimal task allocation on the Cloud-Edge continuum. Throughout the project, an end-to-end example, involving procedural code, performance analysis, and optimal scheduling, will be developed to demonstrate the feasibility, applicability, and potential of the proposed approach.

Methodologies for the Parallelization, Performance Evaluation and Scheduling of Applications for the Cloud-Edge Continuum

Esposito A.
;
Aversa R.;Di Martino B.;Venticinque S.;
2024

Abstract

This paper outlines a project, started in October 2023, and entitled Methodologies for the Parallelization, Performance Evaluation and Scheduling of Applications for the Cloud-Edge Continuum (MPESACC), aimed at developing a holistic approach to parallelize sequential code across the Cloud-Edge Continuum. MPESACC entails the formulation of a Directive-based programming model, integration of performance analysis, bottleneck identification, scheduling, and optimization techniques for the execution of parallelized code. The proposed model, tailored for Cloud-Edge computing, involves the design of a Compiler software capable of decomposing serial codes, annotated with directives, into distributed components for remote execution. Parallelization techniques conveyed through Parallel Patterns and communication templates in the form of Code Skeletons, simplify transformations and adhere to best practices in code parallelization and distribution. The performance analysis models consider crucial aspects of the Cloud-Edge continuum, such as bandwidth, processing capacity, and energy constraints, providing essential feedback for directive and pattern selection. Scheduling considerations address resource availability and computational needs, optimizing time, energy, storage, and bandwidth constraints. Robust mathematical models support scheduling optimization to prevent task failure or idleness due to temporary resource shortages. The proposed Directives guide scheduling algorithms by providing execution constraints, ensuring informed decisions for optimal task allocation on the Cloud-Edge continuum. Throughout the project, an end-to-end example, involving procedural code, performance analysis, and optimal scheduling, will be developed to demonstrate the feasibility, applicability, and potential of the proposed approach.
2024
Esposito, A.; Aversa, R.; Barbierato, E.; Calzarossa, M. C.; Di Martino, B.; Massari, L.; Mongiardo, I. G.; Tessera, D.; Venticinque, S.; Zanussi, L.;...espandi
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11591/526410
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 3
  • ???jsp.display-item.citation.isi??? ND
social impact