Latency is becoming a significant factor in many Internet applications such as P2P sharing and online
gaming. Coupled with the fact that an increasing number of people are using online services for backup
and replication purposes and it is clear that congestion increases exponentially on the network. One of
the ways in which the latency problem can be solved is to remove core network congestion or to limit
it in such a way that it does not pose a problem. In South Africa, Telkom rolled out MSAN cabinets as
part of their Fibre-to-the-curb (FTTC) upgrades. This created an unique opportunity to provide new
services, like BaRaaS, by implementing micro data centres within the MSAN to reduce congestion on
the core network.
It is important to have background knowledge on what exactly latency is and what causes it on a
network. It is also essential to have an understanding of how congestion (and thus latency) can
be avoided on a network. The background literature covered helps to determine which tools are
available to do this, as well as to highlight any possible gaps that exist for new congestion control
mechanisms. A simulation study was performed to determine whether implementing micro data centres inside the
MSAN will in fact reduce latency. Simulations must be done as realistically as possible to ensure that
the results can be correlated to a real-world problem. Two different simulations were performed to
model the behaviour of the network when backup and replication data is sent to the Internet and when
it is sent to a local MSAN. In both models the core network throughput as well as the Round Trip
Times (RTTs) from the client to the Internet and the MSAN cabinets, were recorded. The RTT results
were then used to determine whether latency had been reduced.
Once it was established that micro data centres will indeed help in reducing congestion and latency
on the network, the design of a storage server, for inclusion inside the MSAN cabinet, was done. A
cost benefit analysis was also performed to ensure that the project will be financially viable in the long
term. The cost analysis took into account all the costs associated with the project and then expanded
them over a certain period of time to determine initial expenses. Extra information was then taken into
consideration to determine the possible income per year as well as extra expenditure. It was found that
the inclusion of a micro data centre reduces latency on the core network due to the removal of large
backup data traffic from the core network, which reduces congestion and improves latency. From the
Cost Benefit Analysis (CBA) it was found that the BaRaaS service is viable from a subscription point
of view.
Finally, the relevant conclusions with regard to the effects of data centres in MSAN cabinets on latency
and services were drawn. / Vertraagtyd word 'n belangrike faktor in baie Internet toepassings soos P2P-deel en aanlyn-speletjies.
Gekoppel met die feit dat 'n toenemende getal mense internetdienste gebruik vir rugsteun en replisering,
word opeenhoping in die datanetwerk eksponensieel verhoog. Een van die maniere waarop die
vertraagtydsprobleem opgelos kan word, is om opeenhoping in die kern-datanetwerk te verwyder of
om dit op so 'n manier te beperk dat dit nie 'n probleem veroorsaak nie. In Suid Afrika het Telkom
MSAN-kaste uitgerol as deel van hulle "Fibre-to-the-Curb" (FTTC) opgraderings. Dit het 'n unieke
geleentheid geskep om nuwe dienste te skep, soos BaRaaS, deur mikro-datasentrums in die MSAN-kas
te implementeer om opeenhoping in die kernnetwerk te verminder.
Dit is belangrik om agtergrondkennis te hê van presies wat vertraagtyd is en waardeur dit op die
netwerk veroorsaak word. Dit is ook belangrik om 'n begrip te hê van hoe opeenhoping (en dus
vertraagtyd) op die netwerk vermy kan word. Die agtergrondsliteratuur wat gedek is help om te bepaal
watter instrumente beskikbaar is, asook om moontlikhede na vore te bring vir nuwe meganismes om
opeenhoping te beheer. 'n Simulasiestudie is uitgevoer om vas te stel of die insluiting van datasentrums in die MSAN-kaste
inderdaad 'n verskil sal maak aan die vertraagtyd in die datanetwerk. Twee simulasies is uitgevoer om
die gedrag van die netwerk te modelleer wanneer rugsteun- en repliseringsdata na onderskeidelik die
Internet en die plaaslike MSAN gestuur word. In altwee is die deurset van die kernnetwerk sowel as die
sogenaamde Round Trip Times (RTTs) van die kliënt na die Internet en die MSAN-kaste aangeteken.
Die RTTs-resultate sal gebruik word om te bepaal of vertraagtyd verminder is.
Nadat dit bepaal is dat mikro-datasentrums wel die opeenhoping in die netwerk sal verminder, is
die ontwerp van 'n stoorbediener gedoen, vir insluiting in die MSAN-kas. 'n Koste-ontleding neem
alle koste wat met die projek verband hou in ag en versprei dit dan oor 'n bepaalde tydperk om
die aanvanklike kostes te bepaal. Verdere inligting word voorts in ag geneem om die moontlike
inkomste per jaar sowel as addisionele uitgawes te bepaal. Daar is bevind dat die insluiting van
'n mikro-datasentrum vertraagtyd verminder deur groot rugsteen-dataverkeer van die kernnetwerk
af te verwyder. Die koste-ontleding het gewys dat uit 'n subskripsie-oogpunt, die BaRaaS diens
lewensvatbaar is.
Uiteindelik word relevante gevoltrekkings gemaak oor die effek van datasentrums in MSAN-kaste op
vertraagtyd en dienste. / Dissertation (MEng)--University of Pretoria, 2017. / Electrical, Electronic and Computer Engineering / MEng / Unrestricted
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:up/oai:repository.up.ac.za:2263/61342 |
| Date | January 2017 |
| Creators | van Wyk, David |
| Contributors | Van Wyk, J.H. (Jacques Herman), vanwyk.dvw@gmail.com |
| Publisher | University of Pretoria |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
| Type | Dissertation |
| Rights | © 2017 University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria. |
Page generated in 0.0026 seconds