SABER UCV >
2) Tesis >
Pregrado >

Por favor, use este identificador para citar o enlazar este ítem: http://hdl.handle.net/123456789/748

Título : Proposition and simulation study of a novel heavy hitter detection mechanism
Autor : Cárdenas A., Miguel Á.
Palabras clave : Heavy Hitter
Traffic Engineering
Queue
Fecha de publicación : 10-Nov-2011
Citación : CD TESIS;I2008 C266
Resumen : In the present dissertation is proposed a new mechanism with three (3) implementable variations (counting schemes) for the detection of the most significant so called Heavy Hitters (flows whose load inside a flow-set is very relevant) and or a group of them. The mechanisms is to be applied in egress router queues; it consist in evidencing every certain period of time (called HCP) what is considered as the HHs applying some counting scheme. Moreover, inside each HCP is done a fix number of consecutive queue checks separated by a time interval; at every check, is count the number of packets for each of the found flows and the information of all those checks is related by the counting scheme; that information is only valid during one HCP. The counting schemes are: (1) cumulative, (2) vector, (3) mean. The mechanism is evaluated in a controlled simulation scenario in where can be found TCP and UDP flows with a variety of bit-rates. In particular is denoted the attention to the cumulative scheme which in implementation considerations shows to the fastest and less memory consuming of the schemes. In the study was determined that increasing the number of queue checks improves the selection of the HHs by the mechanism; on the other side, increasing the time interval between checks has not much effect in the selection. Where compared the results of the Cumulative-CS with those offered by the Mean-CS and was found that the performance in finding the most relevant flows was better for the cumulative counting scheme which is also the fastest and less memory consuming of the schemes.
URI : http://saber.ucv.ve/123456789/748
Aparece en las colecciones: Pregrado

Ficheros en este ítem:

Fichero Descripción Tamaño Formato
tesis_version_UCV.pdf1.01 MBAdobe PDFVisualizar/Abrir
Anexo_Resumen_en_español.pdf6.13 kBAdobe PDFVisualizar/Abrir
Anexo_DropTailQueueExplorableGuide.pdf127.46 kBAdobe PDFVisualizar/Abrir

Los ítems de DSpace están protegidos por copyright, con todos los derechos reservados, a menos que se indique lo contrario.

 

Valid XHTML 1.0! DSpace Software Copyright © 2002-2008 MIT and Hewlett-Packard - Comentarios