Persico, Valerio (2016) Network monitoring in public clouds: issues, methodologies, and applications. [Tesi di dottorato]

[img]
Preview
Text
persico_valerio_28.pdf

Download (5MB) | Preview
[error in script] [error in script]
Item Type: Tesi di dottorato
Lingua: English
Title: Network monitoring in public clouds: issues, methodologies, and applications
Creators:
CreatorsEmail
Persico, Valeriovalerio.persico@unina.it
Date: 31 March 2016
Number of Pages: 207
Institution: Università degli Studi di Napoli Federico II
Department: Ingegneria Elettrica e delle Tecnologie dell'Informazione
Scuola di dottorato: Ingegneria dell'informazione
Dottorato: Ingegneria informatica ed automatica
Ciclo di dottorato: 28
Coordinatore del Corso di dottorato:
nomeemail
Garofalo, Francescofrancesco.garofalo@unina.it
Tutor:
nomeemail
Pescapè, AntonioUNSPECIFIED
Date: 31 March 2016
Number of Pages: 207
Uncontrolled Keywords: cloud network; monitoring; cloud
Settori scientifico-disciplinari del MIUR: Area 09 - Ingegneria industriale e dell'informazione > ING-INF/05 - Sistemi di elaborazione delle informazioni
Date Deposited: 03 May 2016 13:04
Last Modified: 17 May 2017 01:00
URI: http://www.fedoa.unina.it/id/eprint/10969

Abstract

Cloud computing adoption is rapidly growing thanks to the carried large technical and economical advantages. Its effects can be observed also looking at the fast increase of cloud traffic: in accordance with recent forecasts, more than 75\% of the overall datacenter traffic will be cloud traffic by 2018. Accordingly, huge investments have been made by providers in network infrastructures. Networks of geographically distributed datacenters have been built, which require efficient and accurate monitoring activities to be operated. However, providers rarely expose information about the state of cloud networks or their design, and seldom make promises about their performance. In this scenario, cloud customers therefore have to cope with performance unpredictability in spite of the primary role played by the network. Indeed, according to the deployment practices adopted and the functional separation of the application layers often implemented, the network heavily influences the performance of the cloud services, also impacting costs and revenues. In this thesis cloud networks are investigated enforcing non-cooperative approaches, i.e.~that do not require access to any information restricted to entities involved in the cloud service provision. A platform to monitor cloud networks from the point of view of the customer is presented. Such a platform enables general customers---even those with limited expertise in the configuration and the management of cloud resources---to obtain valuable information about the state of the cloud network, according to a set of factors under their control. A detailed characterization of the cloud network and of its performance is provided, thanks to extensive experimentations performed during the last years on the infrastructures of the two leading cloud providers (Amazon Web Services and Microsoft Azure). The information base gathered by enforcing the proposed approaches allows customers to better understand the characteristics of these complex network infrastructures. Moreover, experimental results are also useful to the provider for understanding the quality of service perceived by customers. By properly interpreting the obtained results, usage guidelines can be devised which allow to enhance the achievable performance and reduce costs. As a particular case study, the thesis also shows how monitoring information can be leveraged by the customer to implement convenient mechanisms to scale cloud resources without any a priori knowledge. More in general, we believe that this thesis provides a better-defined picture of the characteristics of the complex cloud network infrastructures, also providing the scientific community with useful tools for characterizing them in the future.

Actions (login required)

View Item View Item