CREW: Cost and Reliability aware Eagle‐Whale optimiser for service placement in Fog

Abstract

Integration of Internet of Things (IoT) with industries revamps the traditional ways in which industries work. Fog computing extends Cloud services to the vicinity of end users. Fog reduces delays induced by communication with the distant clouds in IoT environments. The resource constrained nature of Fog computing nodes demands an efficient placement policy for deploying applications, or their services. The distributed and heterogeneous features of Fog environments deem it imperative to consider the reliability performance parameter in placement decisions to provide services without interruptions. Increasing reliability leads to an increase in the cost. In this article, we propose a service placement policy which addresses the conflicting criteria of service reliability and monetary cost. A multiobjective optimisation problem is formulated and a novel placement policy, Cost and Reliability‐aware Eagle‐Whale (CREW), is proposed to provide placement decisions ensuring timely service responses. Considering the exponentially large solution space, CREW adopts Eagle strategy based multi‐Whale optimisation for taking placement decisions. We have considered real time microservice applications for validating our approaches, and CREW has been experimentally shown to outperform the existing popular multiobjective meta‐heuristics such as NSGA‐II and MOWOA based placement strategies.

From zero to fog: Efficient engineering of fog‐based Internet of Things applications

“…Consequently, finding valid deployment options becomes exponentially harder with each added component and is infeasible for larger deployments. Tong et al 38 and, to some extent Heintz et al 36 take a similar approach to FogTorch, while 4‐7,35,39‐55 employ a more efficient heuristics approach to solve the formalized optimization problem. Naas et al 56 offer a heuristics‐based solution as well, but place data replicas rather than services, a challenge that is also addressed in References 57,58.…”

Service Placement for Latency Reduction in the Fog Using Application Profiles

“…An additional factor to consider is the metric guiding the placement process. Mostly, the metrics used are based on network aspects such as resource usage [13], [16], [18], [23], power consumption [21], [24], reliability [25], availability [26], QoS [17], [20], and timerelated metrics such as response time [24], [27], delay [21] and latency [14], [15], [22], [23], [28], [29]. In the case of metrics that concern the applications and their services, QoE, which measures the quality as experienced by the user, is explored as metric [30]; however, no element to describe the applications and the behavior of their services is used during the placement process, except one work [15] that explores the placement of applications according to their requests, but do not categorize the applications accordingly.…”

“…In the case of metrics that concern the applications and their services, QoE, which measures the quality as experienced by the user, is explored as metric [30]; however, no element to describe the applications and the behavior of their services is used during the placement process, except one work [15] that explores the placement of applications according to their requests, but do not categorize the applications accordingly. For the validation, most works use simulation, varying the tool used, largely iFogSim [13], [18], [25], [28], [30], but also YAFS [26], FogTorch [20], Matlab [14], and Python [17], [23]. There is also evaluation via testbed using dockers and containers [16], [24].…”

Towards Auction-Based Function Placement in Serverless Fog Platforms

“…On the one hand, there are static solutions using upfront best practices [103,52,71,112], simulation [60,59,51,67,23,116,129,107,39,115,46], testbed evaluation [55,84,32,37,84,32,10,35,7,78], a combination of those [105,110], or formalized assignment problems [34,24,25,92,47,5,61,72,126,134,65]. On the other hand, dynamic approaches using centralized schedulers [14,121,81,36,6,91,87,135,98,88,125,…”