Energy efficient redundant configurations for real-time parallel reliable servers

Abstract

Modular redundancy and temporal redundancy are traditional techniques to increase system reliability. In addition to being used as temporal redundancy, with technology advancements, slack time in a system can also be used by energy management schemes to save energy. In this paper, we consider the combination of modular and temporal redundancy to achieve energy efficient reliable real-time service provided by multiple servers. We first propose an efficient adaptive parallel recovery scheme that appropriately processes service requests in parallel to increase the number of faults that can be tolerated and thus system reliability. Then we explore schemes to determine the optimal redundant configurations of the parallel servers to minimize system energy consumption for a given reliability goal or to maximize system reliability for a given energy budget. Our analysis results show that small requests, optimistic approaches, and parallel recovery favor lower levels of modular redundancy, while large requests, pessimistic approaches and restricted serial recovery favor higher levels of modular redundancy.

Two-Phase Low-Energy N-Modular Redundancy for Hard Real-Time Multi-Core Systems

“…The main reason to choose the technique of [5] for the comparison is that it is a recent work with similar conditions to the proposed technique, e.g., hard real-time constraints, the use of DVS, and the frame-based application model with task precedence constraints (a set of dependent tasks with a global deadline) running on multi-core platforms. Also, it is noteworthy that for many of the previous works it is not meaningful to compare them with the proposed technique because they considerably differ from ours in application model (e.g., chain of dependent tasks in [6], single-task frame in [13], periodic tasks in [11], [17], and independent tasks in [18]). …”

“…[5] [16], [17], [18], [38], have proposed energy-management techniques for task-level redundancy in multi-core systems. [5] and [16] have considered only one faulty execution for each task to preserve the original system reliability, while for many applications (e.g., the applications that are used in harsh environments) a high level of reliability cannot be achieved unless tolerating multiple faulty tasks [9], [10], [17], [18]. Some works have considered different application models, e.g.…”

“…Some works have considered different application models, e.g. periodic independent real-time tasks in [17] and [38] and parallel independent applications in [18]. However, these works cannot be applied to tasks with precedence constraints (e.g., task graphs [5], [6], [7]), while we consider hard real-time applications with task precedence constraint and propose a scheduling and energy-management technique for these applications.…”

“…It should be noted that there are various techniques to achieve low-energy fault-tolerance in real-time systems (e.g., [6], [7], [8], [11], [13], [16], [17], [18]) and it is beyond the scope of this paper to compare the proposed technique with all these various techniques. The main reason to choose the technique of [5] for the comparison is that it is a recent work with similar conditions to the proposed technique, e.g., hard real-time constraints, the use of DVS, and the frame-based application model with task precedence constraints (a set of dependent tasks with a global deadline) running on multi-core platforms.…”

Modeling and analyzing the energy consumption of fork‐join‐based task parallel programs

“…T. RAUBER AND G. RÜNGER In the domain of real-time scheduling, many DVFS-based techniques have been considered for utilizing waiting times, see, e.g., [54][55][56]. These approaches are usually based on heuristics and are not based on an analytical model as presented in this work.…”