More and more services are moving to the cloud, attracted by the promise of unlimited resources that are accessible anytime, and are managed by someone else. However, hosting every type of service in large cloud datacenters is not possible or suitable, as some emerging applications have stringent latency or privacy requirements, while also handling huge amounts of data. Therefore, in recent years, a new paradigm has been proposed to address the needs of these applications: the edge computing paradigm. Resources provided at the edge (e.g., for computation and communication) are constrained, hence resource management is of crucial importance.

The incoming load to the edge infrastructure varies both in time and space. Managing the edge infrastructure so that the appropriate resources are available at the required time and location is called orchestrating. This is especially challenging in case of sudden load spikes and when the orchestration impact itself has to be limited. This thesis enables edge computing orchestration with increased resource-awareness by contributing with methods, techniques, and concepts for edge resource management. First, it proposes methods to better understand the edge resource demand. Second, it provides solutions on the supply side for orchestrating edge resources with different characteristics in order to serve edge applications with satisfactory quality of service. Finally, the thesis includes a critical perspective on the paradigm, by considering sustainability challenges.

To understand the demand patterns, the thesis presents a methodology for categorizing the large variety of use cases that are proposed in the literature as potential applications for edge computing. The thesis also proposes methods for characterizing and modeling applications, as well as for gathering traces from real applications and analyzing them. These different approaches are applied to a prototype from a typical edge application domain: Mixed Reality. The important insight here is that application descriptions or models that are not based on a real application may not be giving an accurate picture of the load. This can drive incorrect decisions about what should be done on the supply side and thus waste resources.

Regarding resource supply, the thesis proposes two orchestration frameworks for managing edge resources and successfully dealing with load spikes while avoiding over-provisioning. The first one utilizes mobile edge devices while the second leverages the concept of spare devices. Then, focusing on the request placement part of orchestration, the thesis formalizes it in the case of applications structured as chains of functions (so-called microservices) as an instance of the Traveling Purchaser Problem and solves it using Integer Linear Programming. Two different energy metrics influencing request placement decisions are proposed and evaluated.

Finally, the thesis explores further resource awareness. Sustainability challenges that should be highlighted more within edge computing are collected. Among those related to resource use, the strategy of sufficiency is promoted as a way forward. It involves aiming at only using the needed resources (no more, no less) with a goal of reducing resource usage. Different tools to adopt it are proposed and their use demonstrated through a case study.

Deciding where to handle services and tasks, as well as provisioning an adequate amount of computing resources for this handling, is a main challenge of edge computing systems. Moreover, latency-sensitive services constrain the type and location of edge devices that can provide the needed resources. When available resources are scarce there is a possibility that some resource allocation requests are denied. In this work, we propose the VioLinn system to tackle the joint problems of task placement, service placement, and edge device provisioning. Dealing with latency-sensitive services is achieved through proximityaware algorithms that ensure the tasks are handled close to the end-user. Moreover, the concept of spare edge device is introduced to handle sudden load variations in time and space without having to continuously over-provision. Several spare device selection algorithms are proposed with different cost/performance tradeoffs. Evaluations are performed both in a Kubernetes-based testbed and using simulations and show the benefit of using spare devices for handling localized load spikes with higher quality of service (QoS) and lower computing resource usage. The study of the different algorithms shows that it is possible to achieve this increase in QoS with different tradeoffs against cost and performance.

The emerging field of edge computing is suffering from a lack of representative data to evaluate rapidly introduced new algorithms or techniques. That is a critical issue as this complex paradigm has numerous different use cases which translate into a highly diverse set of workload types.

In this work, within the context of the edge computing activity of SPEC RG Cloud, we continue working towards an edge benchmark by defining high-level workload classes as well as collecting and analyzing traces for three real-world edge applications, which, according to the existing literature, are the representatives of those classes. Moreover, we propose a practical and generic methodology for workload definition and gathering. The traces and gathering tool are provided open-source.

In the analysis of the collected workloads, we detect discrepancies between the literature and the traces obtained, thus highlighting the need for a continuing effort into gathering and providing data from real applications, which can be done using the proposed trace gathering methodology. Additionally, we discuss various insights and future directions that rise to the surface through our analysis.

Information and communication technologies are increasingly pervasive in our everyday lives. Their use has greatly evolved from an ancillary service to a component of all our activities, anytime, anywhere. To this aim, we rely heavily on cloud computing and, more recently, on edge computing. Hence, their contribution (or obstruction) to the sustainability of our society at large is pivotal.

Unfortunately, cloud/edge provisioning has a dark side: it too often prioritizes economic gain over the cost of long-lasting sustainability. Also, sustainability is often absent from the discussions in the cloud/edge research community.

To start the discussion and highlight a number of sustainability shortcomings of the cloud and edge computing paradigms, we carry out an exploratory study involving experts-in-the-field, capture their inputs in the form of so-called unsustainable patterns, and complement them with examples of possible countermeasures.

The results of our study include: (i) the definition of a Pattern Model, (ii) a catalog of unsustainable patterns for the cloud and edge computing paradigms, and (iii) the identification of preliminary countermeasures and takeaways in order to make these two paradigms more sustainable.

n/a

The edge computing paradigm comes with a promise of lower application latency compared to the cloud. Moreover, offloading user device computations to the edge enables running demanding applications on resource-constrained mobile end devices. However, there is a lack of workload models specific to edge offloading using applications as their basis.In this work, we build upon the reconfigurable open-source mixed reality (MR) framework MR-Leo as a vehicle to study resource utilisation and quality of service for a time-critical mobile application that would have to rely on the edge to be widely deployed. We perform experiments to aid estimating the resource footprint and the generated load by MR-Leo, and propose an application model and a statistical workload model for it. The idea is that such empirically-driven models can be the basis of evaluations of edge algorithms within simulation or analytical studies.A comparison with a workload model used in a recent work shows that the computational demand of MR-Leo exhibits very different characteristics from those assumed for MR applications earlier.

The increasing diversity of connected devices leads to new application domains being envisioned. Some of these need ultra low latency or have privacy requirements that cannot be satisfied by the current cloud. By bringing resources closer to the end user, the recent edge computing paradigm aims to enable such applications.

One critical aspect to ensure the successful deployment of the edge computing paradigm is efficient resource management. Indeed, obtaining the needed resources is crucial for the applications using the edge, but the resource picture of this paradigm is complex. First, as opposed to the nearly infinite resources provided by the cloud, the edge devices have finite resources. Moreover, different resource types are required depending on the applications and the devices supplying those resources are very heterogeneous. This thesis studies several challenges towards enabling efficient resource management for edge computing.

The thesis begins by a review of the state-of-the-art research focusing on resource management in the edge computing context. A taxonomy is proposed for providing an overview of the current research and identify areas in need of further work.

One of the identified challenges is studying the resource supply organization in the case where a mix of mobile and stationary devices is used to provide the edge resources. The ORCH framework is proposed as a means to orchestrate this edge device mix. The evaluation performed in a simulator shows that this combination of devices enables higher quality of service for latency-critical tasks.

Another area is understanding the resource demand side. The thesis presents a study of the workload of a killer application for edge computing: mixed reality. The MR-Leo prototype is designed and used as a vehicle to understand the end-to-end latency, the throughput, and the characteristics of the workload for this type of application. A method for modeling the workload of an application is devised and applied to MR-Leo in order to obtain a synthetic workload exhibiting the same characteristics, which can be used in further studies. 

In the emerging edge computing architecture, several types of devices have computational resources available. In order to make efficient use of those resources, deciding on which device a task should execute is of great importance. Existing works on task placement in edge computing focus on a resource supply side consisting of stationary devices only. In this paper, we consider the addition of mobile edge devices. We explore how mobile and stationary edge devices can augment the original task placement problem with a second placement problem: the placement of the mobile edge devices. We propose the ORCH framework in order to solve the joint problem in a distributed manner and evaluate it in the context of a spatially-changing load. Our implementation of the combined task and edge placement algorithms shows a normalized 83% delay-sensitive task completion rate compared to a perfect edge placement strategy.

Edge computing is a recent paradigm where computing resources are placed close to the user, at the edge of the network. This is a promising enabler for applications that are too resource-intensive to be run on an end device, but at the same time require too low latency to be run in a cloud, such as for example mixed reality (MR).

In this work, we present MR-Leo, a prototype for creating an MR-enhanced video stream. It enables offloading of the point cloud creation and graphic rendering at the edge. We study the performance of the prototype with regards to latency and throughput in five different configurations with different alternatives for the transport protocol, the video compression format and the end/edge devices used.

The evaluations show that UDP and MJPEG are good candidates for achieving acceptable latency and that the design of the communication protocol is critical for offloading video stream analysis to the edge.

The edge computing paradigm has recently attracted research efforts coming from different application domains. However, evaluating an edge platform or algorithm is impeded by the lack of suitable benchmarks. We propose a methodology for characterizing edge workloads from different application domains. It is a first step towards defining workloads to be included in a future edge benchmarking suite. We evaluate the methodology on three use cases and find that defining a common and standard set of workloads is plausible.