Paulo A. L. Rego

Elasticity in cloud computing: a survey

Cloud computing is now a well-consolidated paradigm for on-demand services provisioning on a pay-as-you-go model. Elasticity, one of the major benefits required for this computing model, is the ability to add and remove resources “on the fly” to handle the load variation. Although many works in literature have surveyed cloud computing and its features, there is a lack of a detailed analysis about elasticity for the cloud. As an attempt to fill this gap, we propose this survey on cloud computing elasticity based on an adaptation of a classic systematic review. We address different aspects of elasticity, such as definitions, metrics and tools for measuring, evaluation of the elasticity, and existing solutions. Finally, we present some open issues and future directions. To the best of our knowledge, this is the first study on cloud computing elasticity using a systematic review approach.

FairCPU: Architecture for Allocation of Virtual Machines Using Processing Features

This paper proposes an architecture to handle the allocation of virtual machines based on the processing power for heterogeneous Clouds, where there is a wide variety of CPU types. Our major contribution is a novel representation of the processing capacity in terms of the Processing Unit (PU) and the CPU usage limitation in order to isolate the processing capability from the Physical Machine (PM) where the Virtual Machine (VM) is allocated. The efficiency of the proposed architecture is validated by extensive replications of five experiments using a real private cloud. The results show that it is possible to use the proposed idea to define a PU, supported by the CPU usage limitation, to enable the VMs processing power remain at the same level regardless of the PM.

MpOS: a multiplatform offloading system

Mobile applications and services have changed different aspects of modern life, besides allowing to be accessed by mobile devices at any moment, regardless of time and place. These devices usually interact with more powerful machines usually hosted at Internet on public clouds. This paper presents MpOS (Multiplatform Offloading System), a framework that supports a method-based offloading technique for applications of different mobile platforms, and was also developed initially for Android and Windows Phone. The framework provides several services to support the offloading process (e.g., discovery service, deployment service, and network profiler). In order to assess the proposed framework, we first developed an application using the MpOS framework for both Android and Windows Phone platforms. Finally, we performed an experiment to evaluate the developed applications performance. The result shows that offloading operation achieved a speedup of 14x running on cloudlet when compared with the execution on the mobile device.

Performing computation offloading on multiple platforms

An offloading framework designed for supporting multiple platforms is proposed.The solution supports Android and Windows Phone mobile applications.Developers can use static or dynamic offloading decision.The offloading technique improves the performance of mobile applications.The type of serialization used impacts the offloading performance. Mobile devices such as smart phones and tablets are increasingly important tools in daily routine. These devices generally interact with more powerful machines usually hosted on public clouds. In this context, this paper presents MpOS (Multiplatform Offloading System), a framework that supports a method-based offloading technique for applications of different mobile platforms (Android and Windows Phone). In addition, details of MpOS main components and services as well as code examples are presented. To evaluate the proposed solution and to analyse the impact of different serialization types on the offloading performance, we developed two applications and performed several experiments on both Android and Windows Phone platforms using WiFi and 4G/LTE connections to access the remote execution environments. Our results state that offloading to a cloudlet has provided the best performance for both Android and Windows Phone platforms, beyond showing that the type of serialization used by the framework directly impacts on the offloading performance.

An OpenFlow-Based Elastic Solution for Cloud-CDN Video Streaming Service

Media streaming services are responsible for the significant increase of Internet traffic. This fact generates the need for better computing resources management in order to maintain such services always available. Cloud computing offers an elastic infrastructure that provides computing resources on demand in order to maintain such services always available. Moreover, emerging network architectures, such as Software Defined Networks, provide mechanisms that simplify the implementation of advanced network functions. In this context, this paper proposes an elastic approach to manage a video streaming service. The solution includes an OpenFlow-based load balancer, and an elasticity management mechanism that increases or decreases the amount of active streaming servers, considering customers demand. A prototype was developed using a private cloud infrastructure and experiments were performed to show the viability and effectiveness of our solution, in terms of load balancing and elasticity functionality.

MapReduce performance evaluation for knowledge-based recommendation of context-tagged photos

Recommendation systems are a subclass of information filtering systems that aims at helping users in retrieving information. Recently, contextual information proved to be effective in improving the quality of results of Recommender Systems. However, Context-aware Recommender Systems still suffer performance issues for real-time recommendation, mainly due to the amount of items that should be considered for recommendation. In this paper, we present an evaluation of using MapReduce and its integration with a mobile system for implementing a knowledge-based algorithm for context-aware recommendation. To be effective, this photo recommendation algorithm should work with a large set of images annotated with contextual information. The MapReduce algorithm parallelizes the processing required to generate the recommendation results and so improved the system performance. The results of performance analysis showed, for instance, that cloud-based version of the reccomendation reaches a speedup of 7x with a image base with more than 41 million photos.

Decision Tree-Based Approaches for Handling Offloading Decisions and Performing Adaptive Monitoring in MCC Systems

Mobile cloud computing (MCC) has emerged as a solution to overcome the resource constraints of mobile devices by using computation offloading to execute mobile application tasks on remote servers, thus enhancing performance and reducing the energy consumption of mobile devices. Nevertheless, the effectiveness of an offloading solution is determined by its ability to infer when offloading will improve performance. In this context, several solutions have been proposed to handle computational offloading operations and the decisions of when and where to offload. The problem is that such decisions depend on periodic monitoring of several metrics and usually involve compute intensive task that, when executed on mobile devices, can contribute to overhead the system. Thus, this paper proposes a novel approach for handling offloading decisions using decision trees and an adaptive monitoring scheme that allows MCC systems to monitor only the metrics that are relevant to the offloading decision. The results show that computation offloading can be beneficial for improving the performance of mobile applications and the energy consumption of mobile devices can be reduced by using the proposed adaptive monitoring scheme.

An architecture for providing elasticity based on autonomic computing concepts

Elasticity is a feature quite important for cloud computing and it is related to how a system autonomously adapts its capacity over time to fit the workload variation. In this context, this paper proposes an elastic architecture for cloud computing based on autonomic computing concepts, such as control loops and thresholds-based rules. In order to validate the proposed solution, we designed two experiments that use microbenchmarks on private and hybrid cloud environments. The results show cloud computing and autonomic computing may be leveraged together for elasticity provisioning.

Using Mobile Cloud Computing for Developing Context-Aware Multimedia Applications: A Case Study of the CAOS Framework

Mobile cloud computing (MCC) and context-aware computing (CAC) are research topics in growing evidence. The former seeks to leverage cloud computing features to improve the performance of mobile applications and reduce the energy consumption of mobile devices, while the latter seeks effective ways to build applications that react to changes in its context environment. This short-course aims at presenting main concepts, solutions, and technologies related to the integration of MCC and context-aware applications. We will present different motivational scenarios, examples of applications, as well as a practical guide to the development of a context-aware multimedia Android application using the framework CAOS. In addition, we will highlight research challenges and opportunities that come with such integration.

Evaluating the elasticity of multimedia applications in a cloud computing environment using network metrics

Actually, Internet users have a broad and varied range of possible services to access, such as enterprise applications and entertainment. These applications are increasingly generating a lot of network traffic mainly due to multimedia streaming. Cloud computing and its elasticity capability are some of the reason for such increase in multimedia traffic, once new applications and services are easily deployed in cloud environments. However, the way to measure and evaluate the elasticity is still quite varied and can be considered an open issue. In this context, this study proposes a set of elasticity metrics based on network data for evaluating multimedia applications performance. Our main contributions are: (i) metrics for elasticity evaluation in computational cloud based on network metrics; and (ii) two experiments in a private cloud, using workloads generated by benchmark to evaluate the elasticity and validating the proposed metrics.