Multi-Access Edge Computing (MEC) is increasingly being adopted as the de facto enabler for ultra-low latency access to application services. By placing application services on MEC servers situated in proximity to end users, MEC avoids the large network latencies frequently experienced while accessing cloud services. MEC is envisioned as the fundamental enabler for a number of ultra-low latency safety-critical systems, including data inferencing for autonomous vehicles amongst others. The MEC paradigm is, however, highly susceptible to various types of faults such as MEC server downtime, communication link faults, network hardware faults and so on owing to the heterogeneity of hardware configurations and diverse geographies of operations. For real-time and safety-critical workloads, averting the impact of faults is a key facet. To address this challenge, we synthesize a fault classification policy for MEC that categorizes a fault as critical requiring immediate rectification or non-critical by leveraging Probabilistic Model Checking, a Formal Methods technique, to ensure probabilistic guarantees with respect to a specified failure context. We present experimental results on a real-world datasets to show the effectiveness of our approach.

In Multi-Access Edge Computing (MEC), a number of mechanisms exist to determine the optimal placement of monolithic service workflows. For applications designed as microservice workflow architectures, service placement schemes need to be revisited owing to the inherent interdependencies which exist between microservices. The dynamic environment, with stochastic user movement and service invocations, along with a large placement configuration space makes microservice placement in MEC a challenging task. Additionally, owing to user mobility, a placement scheme may need to be recalibrated, triggering service migrations to maintain the advantages offered by MEC. Existing microservice placement and migration schemes consider on-demand strategies. In this work, we take a different route and propose a Reinforcement Learning (RL) based proactive mechanism using a Learning Automata (LA) for microservice placement and migration that on one hand, keeps track of user mobility and resorts to migration when necessary, while on the other hand, keeps track of server residual capacities so that no server is overloaded. We use the San Francisco Taxi dataset to validate our approach. Experimental results show the effectiveness of our approach in comparison to other methods.

Multi-Access Edge Computing (MEC) is increasingly growing in prominence as the de facto enabler for ultra-low latency access to services. MEC averts the high network latencies often encountered in accessing cloud services by deploying application instances on edge servers situated near Internet-of-Things (IoT) device users. Workloads generated by IoT devices can then either be executed locally on the devices or offloaded to the MEC servers. A key cornerstone of the MEC environment is a service placement policy that determines the deployment of services on MEC servers. A service placement policy plays a critical role towards determining the trade-offs involved between latency experienced by users as a function of the resource contention and the resulting energy consumption. In this context, we propose a static-dynamic service placement policy for MEC. The static policy is geared towards placement of services in a prioritised order by leveraging Probabilistic Model Checking, a Formal Methods technique, to ensure probabilistic guarantees on the trade-offs between latencies and energy consumption of edge sites. The dynamic policy alters the static service allocation to cater to runtime variability in latency requirements. We present experimental results on a real-world service usage dataset to show the benefits of our approach over conventional approaches.

Sustainable computing is gaining priority for data centers due to need for mandatory compliance with carbon emission reporting regulations. Hence, accurately estimating the energy consumption of servers in data centers has become quintessential. Apart from estimating carbon emissions at the server level, quantifying carbon emissions at the virtualization layer is also crucial because it provides a more precise and detailed understanding of the possible hotspots and mismatched utilization which can help in taking remedial actions. In this paper, we take a first step towards modeling energy consumption at the Logical PARtitions (LPARs) level of data centers driven by IBM POWER9 Systems. We experimentally validate our approach on utilization metrics from the data center of the CIO office of IBM and compare it with the instrumented energy measurements wherein we demonstrate on average 90–95 % prediction accuracy for our model.

The advent of Multi-Access Edge Computing (MEC) has enabled service providers to mitigate high network latencies often encountered in accessing cloud services by deploying containerized application instances on edge servers situated near end users. MEC servers are, however, susceptible to various types of failures such as communication link failures, hardware failures and so on. A fault recovery strategy determines which MEC servers to utilize to re-deploy application containers in the event of a failure. In this work, we propose a two-fold fault recovery strategy characterized by application priority. We propose a Formal Methods driven local recovery strategy for high-priority applications. We use Stochastic Multi-Player Games as a Formal Model to characterize the interactions between the different components in an MEC environment. We use objectives specified in Probabilistic Alternating-Time Temporal Logic with a Probabilistic Model Checker to derive recovery strategies considering all possible execution scenarios of the model. For lower priority applications, we resort to a global recovery strategy by designing a greedy heuristic considering each server’s failure probability. We use benchmark datasets to validate our approach. Experimental results show an average 14% reduction in latency with our approach in comparison with other state-of-the-art methods.

With the rapid proliferation of strategic alliances between service providers, enterprises cooperate towards service quality improvement and provide lower cost service bundles. This article presents a novel solution to the minimum cost service bundle selection problem for workflows in the presence of singleton subscription costs and service bundle offerings and compatibility requirements. Given a workflow specifying a set of tasks and a set of candidate services for each task, with a set of compatibility constraints between services, the selection problem has the objective of selecting the most suitable service offering(s) for each task. In this article, we analyze the selection problem in the presence of service bundle offerings. We present a novel multi-partite hyper-graph visualization of the selection problem and analyze its hardness. Additionally we present a novel combination of ILP and abstraction refinement as a potential solution, that is shown to expedite a naïve ILP based solution. We present experiments to substantiate this claim.

Multi-Access Edge Computing (MEC) has emerged as a promising new paradigm allowing low latency access to services deployed on edge servers to avert network latencies often encountered in accessing cloud services. A key component of the MEC environment is an auto-scaling policy which is used to decide the overall management and scaling of container instances corresponding to individual services deployed on MEC servers to cater to traffic fluctuations. In this work, we propose a Safe Reinforcement Learning (RL)-based auto-scaling policy agent that can efficiently adapt to traffic variations to ensure adherence to service specific latency requirements. We model the MEC environment using a Markov Decision Process (MDP). We demonstrate how latency requirements can be formally expressed in Linear Temporal Logic (LTL). The LTL specification acts as a guide to the policy agent to automatically learn auto-scaling decisions that maximize the probability of satisfying the LTL formula. We introduce a quantitative reward mechanism based on the LTL formula to tailor service specific latency requirements. We prove that our reward mechanism ensures convergence of standard Safe-RL approaches. We present experimental results in practical scenarios on a test-bed setup with real-world benchmark applications to show the effectiveness of our approach in comparison to other state-of-the-art methods in literature. Furthermore, we perform extensive simulated experiments to demonstrate the effectiveness of our approach in large scale scenarios.

In recent times, Multi-Access Edge Computing (MEC) has emerged as a new paradigm allowing low-latency access to services deployed on edge nodes offering computation, storage and communication facilities. Vendors deploy their services on MEC servers to improve performance and mitigate network latencies often encountered in accessing cloud services. An allocation policy determines how to allocate service requests from users to MEC servers. A number of proposals for binding user service requests to nearby edge servers enroute have been proposed in literature. However, none of these proposals, to the best of our knowledge, provide quantitative guarantees on performance metrics. Indeed, the evolving environment, along with a large allocation configuration space makes proving performance guarantees for such allocation policies a challenging task. Further, the implications of MEC server failures on allocation policies have been relatively unexplored. To address such issues, we propose a trace driven approach to derive a formal model of allocation policies and perform quantitative verification to produce probabilistic guarantees on performance metrics. We use the San Francisco taxi dataset, the LDNS availability dataset and allocation policies from recent literature to validate our approach. Experimental results demonstrate how our model can be utilized to quantitatively compare performance metrics of service allocation policies in MEC systems.

In recent times, Mobile Edge Computing (MEC) has emerged as a new paradigm allowing low-latency access to services deployed on edge nodes offering computation, storage and communication facilities. Vendors deploy their services on MEC servers to improve performance and mitigate network latencies often encountered in accessing cloud services. An allocation policy determines how to allocate service requests from mobile users to MEC servers. A number of proposals for binding user service requests to nearby edge servers enroute have been proposed in literature. However, none of these proposals, to the best of our knowledge, provide quantitative performance guarantees on the quality of service metrics. Indeed, the evolving environment, along with a large allocation configuration space makes proving performance guarantees for such allocation policies a challenging task. To address such issues, we propose a trace driven approach to derive a formal model of allocation policies and perform quantitative verification to produce probabilistic guarantees on performance metrics. We use benchmark real world MEC server and user datasets and a mobility aware allocation and migration policy from recent literature to validate our model. Experimental results show our model's effectiveness in quantitatively reasoning about service allocation performance metrics in MEC systems.

In recent times, Mobile Edge Computing (MEC) has emerged as a new paradigm allowing low-latency access to services deployed on edge nodes offering computation, storage and communication facilities. Vendors deploy their services on MEC servers to improve performance and mitigate network latencies often encountered in accessing cloud services. A service placement policy determines which services are deployed on which MEC servers. A number of mechanisms exist in literature to determine the optimal placement of services considering different performance metrics. However, for applications designed as microservice workflow architectures, service placement schemes need to be re-examined through a different lens owing to the inherent interdependencies which exist between microservices. Indeed, the dynamic environment, with stochastic user movement and service invocations, along with a large placement configuration space makes microservice placement in MEC a challenging task. Additionally, owing to user mobility, a placement scheme may need to be recalibrated, triggering service migrations to maintain the advantages offered by MEC. Existing microservice placement and migration schemes consider on-demand strategies. In this work, we take a different route and propose a Reinforcement Learning based proactive mechanism for microservice placement and migration. We use the San Francisco Taxi dataset to validate our approach. Experimental results show the effectiveness of our approach in comparison to other state-of-the-art methods.