Marc Manzano

On the Characterization of the Structural Robustness of Data Center Networks

Data centers being an architectural and functional block of cloud computing are integral to the Information and Communication Technology (ICT) sector. Cloud computing is rigorously utilized by various domains, such as agriculture, nuclear science, smart grids, healthcare, and search engines for research, data storage, and analysis. A Data Center Network (DCN) constitutes the communicational backbone of a data center, ascertaining the performance boundaries for cloud infrastructure. The DCN needs to be robust to failures and uncertainties to deliver the required Quality of Service (QoS) level and satisfy Service Level Agreement (SLA). In this paper, we analyze robustness of the state-of-the-art DCNs. Our major contributions are: (a) we present multi-layered graph modeling of various DCNs; (b) we study the classical robustness metrics considering various failure scenarios to perform a comparative analysis; (c) we present the inadequacy of the classical network robustness metrics to appropriately evaluate the DCN robustness; and (d) we propose new procedures to quantify the DCN robustness. Currently, there is no detailed study available centering the DCN robustness. Therefore, we believe that this study will lay a firm foundation for the future DCN robustness research.

An empirical study of Cloud Gaming

Online gaming connects players from all over the world together for fun and entertainment, and has been regarded as one of the most profitable and popular Internet services. Besides, there is a growing trend towards moving local applications to remote data centers: this is often referred to as the cloud. With the purpose of studying the impact of Cloud Gaming on the access network load, in this paper we carry out an empirical network traffic analysis of two well-known cloud gaming platforms: On-Live and Gaikai. Traffic traces have been collected and analysed from five different games of both platforms. Cloud gaming has been observed to be remarkably different from traditional online gaming in terms of network load and traffic characteristics. Moreover, the traces have revealed similarities between the two platforms regarding the packet size distribution, and differences concerning the packet inter-arrival times. However, each platform shows a similar traffic pattern for most of the games it serves. Nonetheless, the racing and shooter games considered in this work demand more bandwidth than other game-genres.

On the Connectivity of Data Center Networks

Data Center Networks (DCNs) constitute the communication backbone for the cloud computing paradigm. Recently, network connectivity analysis in terms of reliability has received attention from the network research community. The traditional network features are useful; however, they are insufficient to determine how well-connected or well-designed a DCN is against the node or link removals. In this letter, we present a connectivity analysis of three well-known DCN architectures, namely: (a) ThreeTier, (b) FatTree, and (c) DCell. Our analysis reveals that the classic connectivity measures are inadequate for evaluating DCN connectivity. Therefore, we propose μ-A2TR, a novel metric to characterize network connectivity in the case of node or link failures. Experimental results reveal that the DCNs exhibit a moderate level of connectivity in the case of random node removals. However, connectivity decays abruptly when considering the targeted nodes removal. Moreover, the connectivity analysis depicts significant differences among the considered DCNs.

Endurance: A new robustness measure for complex networks under multiple failure scenarios

Society is now, more than ever, highly dependent on the large-scale networks that underpin its functions. In relatively recent times, significant failures have occurred on large-scale networks that have a considerable impact upon sizable proportions of the world’s inhabitants. The failure of infrastructure has, in turn, begot a subsequent loss of services supported by that network. Consequently, it is now vitally important to evaluate the robustness of such networks in terms of the services supported by the network in question. Evaluating network robustness is integral to service provisioning and thus any network should include explicit indication of the impact upon service performance. Traditionally, network robustness metrics focused solely on topological characteristics, although some new approaches have considered, to a degree, the services supported by such networks. Several shortcomings of these new metrics have been identified. With the purpose of solving the drawbacks of these metrics, this paper presents a new measure called endurance, which quantifies the level of robustness supported by a specific topology under different types of multiple failure scenarios, giving higher importance to perturbations affecting low percentages of elements of a network. In this paper, endurance of six synthetic complex networks is computed for a range of defined multiple failure scenarios, taking into account the connection requests that cannot be satisfied. It is demonstrated that our proposal is able to quantify the robustness of a network under given multiple failure scenarios. Finally, results show that different types of networks react differently depending on the type of multiple failure.

Robustness analysis of real network topologies under multiple failure scenarios

Nowadays the ubiquity of telecommunication networks, which underpin and fulfill key aspects of modern day living, is taken for granted. Significant large-scale failures have occurred in the last years affecting telecommunication networks. Traditionally, network robustness analysis has been focused on topological characteristics. Recently approaches also consider the services supported by such networks. In this paper we carry out a robustness analysis of five real backbone telecommunication networks under defined multiple failure scenarios, taking into account the consequences of the loss of established connections. Results show which networks are more robust in response to a specific type of failure.

A multiple failure propagation model in GMPLS-based networks

In this article, a new model to simulate different failure propagation scenarios in GMPLS-based networks is proposed. Several types of failures and malfunctions may spread along the network following different patterns (hardware failures, natural disasters, accidents, configuration errors, viruses, software bugs, etc.). The current literature presents several models for the spreading of failures in general networks. In communication networks, a failure affects not only nodes but also the connections passing through those nodes. The model in this article takes into account GMPLS node failures, affecting both data and control planes. The model is tested by simulation using different types of network topologies. In addition, a new method for the classification of network robustness is also introduced.

Dissecting the protocol and network traffic of the OnLive cloud gaming platform

Cloud gaming is a new paradigm that is envisaged to play a pivotal role in the video game industry in forthcoming years. Cloud gaming, or gaming on demand, is a type of online gaming that allows on-demand streaming of game content onto non-specialised devices (e.g. PC, smart TV, etc.). This approach requires no downloads or game installation because the actual game is executed on the game company’s server and is streamed directly to the client. Nonetheless, this revolutionary approach significantly affects the network load generated by online games. As cloud gaming presents new challenges for both network engineers and the research community, both groups need to be fully conversant with these new cloud gaming platforms. The purpose of this paper is to investigate OnLive, one of the most popular cloud gaming platforms. Our key contributions are: (a) a review of the state-of-the-art of cloud gaming; (b) reverse engineering of the OnLive protocol; and (c) a synthetic traffic model for OnLive.

Robustness surfaces of complex networks

Despite the robustness of complex networks has been extensively studied in the last decade, there still lacks a unifying framework able to embrace all the proposed metrics. In the literature there are two open issues related to this gap: (a) how to dimension several metrics to allow their summation and (b) how to weight each of the metrics. In this work we propose a solution for the two aforementioned problems by defining the R*-value and introducing the concept of robustness surface (Ω). The rationale of our proposal is to make use of Principal Component Analysis (PCA). We firstly adjust to 1 the initial robustness of a network. Secondly, we find the most informative robustness metric under a specific failure scenario. Then, we repeat the process for several percentage of failures and different realizations of the failure process. Lastly, we join these values to form the robustness surface, which allows the visual assessment of network robustness variability. Results show that a network presents different robustness surfaces (i.e., dissimilar shapes) depending on the failure scenario and the set of metrics. In addition, the robustness surface allows the robustness of different networks to be compared.

Vulnerability of core networks under different epidemic attacks

This paper is focused on the propagation of failures within a core telecommunication network (i.e. virus or malicious code spreading), which are commonly represented using epidemic models. This work conducts an analysis of the vulnerability of five core telecommunication networks subject to different epidemic attacks. The difference between the attacks relies on the method used to select the initial set of infected nodes. As a novelty, the investigation is focused on the initial stage of the epidemics (first 48 hours). The analysis is carried out taking into account the topological connectivity of the networks. Results are presented in two phases. The first one concerns the first 24 hours of the epidemics evolution and shows the reaction of core networks to different types of epidemic attacks. The second one is focused on the time frame between 24 and 48 hours and presents a study of the repairing times needed to eradicate the epidemics. Results show that the core telecommunication networks react differently depending on the type of epidemics. Thus, different repairing times need to be applied in order to maintain a specific level of connectivity of the networks.

Failure propagation in GMPLS optical rings: CTMC model and performance analysis

Abstract Network reliability and resilience has become a key design parameter for network operators and Internet service providers. These often seek ways to have their networks fully operational for at least 99.999% of the time, regardless of the number and type of failures that may occur in their networks. This article presents a continuous-time Markov chain model to characterise the propagation of failures in optical GMPLS rings. Two types of failures are considered depending on whether they affect only the control plane, or both the control and data planes of the node. Additionally, it is assumed that control failures propagate along the ring infecting neighbouring nodes, as stated by the Susceptible-Infected-Disabled (SID) propagation model taken from epidemic-based propagation models. A few numerical examples are performed to demonstrate that the CTMC model provides a set of guidelines for selecting the appropriate repair rates in order to attain specific availability requirements, both in the control plane and the data plane.