Riverbed Modeler Software 18
The performance of variousnetworks is primarily affected by the nature of applications that areutilized in the process. These forms of stress are likely to changethe working parameters of the requests (WCNA & Zeng, 2016). TheRiverbed modeller software is regarded as one of the fastest discreteevent-simulation engines that are utilized in the analysis anddesigning of the communication networks. This software is composed ofa suite of technologies that exhibit development environment.Therefore, it uses various network types and technologies such asVoIP, TCP, MPLS, OSPFv3, and the IPv6 (Aboulia, 2011).
The riverbed modeller uses theabove system types and techniques to analyze other networks byconducting a comparison analysis about the impact of technologydesigns, as well as end-to-end behaviour. To accomplish this, it letsthe users to test and demonstrate technology design before the actualproduction. Other benefits that can be derived from the use of thissoftware the evaluation of enhancements to standard-based protocolsand increasing the network and research and development productivity.Finally, it facilitates the development of proprietary wirelessprotocols and technologies, therefore promoting an improvement in thetechnological context. It is therefore important to identify ways towhich the networks are affected by loading of different traffic aswell as applications that are likely to exert additional stress onthe requests as this will culminate in the performances of theapplication being substantially affected.
Additionally, there are variousartefacts that can be used to highlight the impact of connectivity innetworks and their corresponding performances. Some of the conceptsdiscussed include bandwidth, packet drops, packet loss, jitter, andbit error rate tester. Each of the available concepts indicates thedeviations from normalcy of operations and how such factors can becorrected. Finally, an analysis of the quality of service highlightsthe overall performance of the computer system.
2.1 Project Title 6
2.2 Academic Question 6
2.3 Aims 6
2.4 Objectives 6
Literature Review 6
Types of Latency 10
Internet Latency 11
WAN Latency 11
Audio Latency 11
Operational Latency 12
Mechanical Latency 12
Computer and Operating System Latency 12
Latency Testing 13
Reducing Latency 13
Packet Drops 15
Packet Loss 17
Error Rate 17
The Bit Error Rate Tester 20
Quality of Service 20
Full details of the Artifact 21
Birmingham Subnet 21
London Subnet 22
The Whole WAN Model 23
Voice Delay 26
Voice jitter 27
DB Query 28
Video Conferencing (overlaid statistics) 32
Video Conferencing (stacked statistics) 33
References and Bibliography 38
A: Birmingham Subnet 40
B: London Subnet 41
C: The Whole WAN Model 42
D: FTP 43
E: Voice Delay 43
F: Voice jitter 44
G: DB Query 44
H: HTTP 45
I: Video Conferencing (overlaid statistics) 45
J: Video Conferencing (stacked statistics) 46
K: Email 47
The applications’ performancesare affected by various factors depending on the application andother network specifications. The study has identified various issuesrelating to Application Performance Management (APM) and how suchchallenges can influence the performance of the applications(Aboulia, 2011) negatively.
The first factor that is likelyto affect the performances of the requests is the complexity of therequests. In the current times, the common trend is the distributionof software components in addition to cloud services to enable thecompletion of complex business services. Based on the fact that thecomponents are required to function concurrently, it becomes tough toachieve the desirable results.
Another factor that should betaken into consideration when analyzing application performances isthe application design. In instances whereby the applications havebeen specified, performance goals should be delineated together withthe details of the environment whereby the application is expected tobe used. Application testing should also be given extensive attentionsince most of the current applications are developed in thesimulation labs without the added effort of testing the requests inreal world situations. There is therefore need to transport suchapplications across highly distributed network architectures for theperformances to be monitored and optimized (Aboulia, 2011).
By using the Riverbed Modelersoftware, this paper will provide an analysis of various factorsaffecting the performance of the applications. To begin with, theriverbed modeller will be used to establish the effects of stress ona network based on different applications. Additionally, the paperwill prove how various systems are affected by loading of differenttraffic (ICIST (Conference), Dregvaite & Damaševičius, 2014).The results will be based on performances analysis conducted on allthe major subnets provided.
Use of Riverbed network modellerto investigate the stress on a network through the use of differentapplications on a WAN.
How is network performanceaffected by the loading of various types of traffic from differentapplications?
• To design and build anadvanced WAN network model in Riverbed.
• To assess the effects onperformance when subjected to various stresses of applications.
• To perform a performanceanalysis of the model under stress subjected by the variousapplication.
• To examine the response ofperformance parameter to application stress and how they relate.
• To determine the effect onQoS (Quality of Service) of the network under these stresses.
• To establish the optimumoperation point of performance under the stress subjected.
In simple terms, bandwidth can bedescribed as the ability of any electronic device to transmitinformation. In this regard, the communication device can be acomputer network (ICIST (Conference), Dregvaite & Damaševičius,2014). The term is also used to describe the range within a band ofwavelengths and frequencies that are occupied by a modulated carrierwave (Sterbenz et. al., 2001). On the other hand, bandwidth refers tothe capacity of data transfer within an electronic communicationsystem. The three major factors that determine the definition ofbandwidth with regards to informational technology depend on a seriesof factors (Iser, Minker & Schmidt, 2008).
Bandwidth is used as a synonymfor “data transfer rate” when discussing computer networks.Therefore, it refers to the amount of data that can be carried fromone point to another within a given period usually given in seconds(Sterbenz et. al., 2001). The level of network bandwidth is oftengiven by bits per second, depicted as bps. Due to technologicaladvancements, modern systems have advanced their operations to thepoint whereby their speeds can be measured in millions of bits persecond. These readings are provided in megabites per second orgigabites per second (Mbps and Gbps respectively) (Iser, Minker &Schmidt, 2008). It is essential to note that the network performanceis affected by various factors and not limited to bandwidth alone(Sterbenz et. al., 2001). These factors include packet loss, latency,and jitter. Because network paths are composed of a series of thesuccession of links, the end-to-end bandwidth is usually limited tothe bandwidth that has the lowest speed, otherwise known as thebottleneck (Holzbecher, 2012).
Figure 1: Bandwidth
The nature of the bandwidths willmassively depend on the application being carried out. For example,in most instances, the instant messaging conversations are likely totake approximately 1000 bits per second (bps) (ICIST (Conference),Dregvaite & Damaševičius, 2014). Sufficient bandwidth refersto the highest level of reliability than can be offered by a pathwith regards to transmission. The bandwidth test is therefore used tomeasure the effective bandwidth. It is done by continuously measuringthe amount of time taken by a particular file to switch from thepoint of origin to the other site after successful downloading (Iser,Minker & Schmidt, 2008).
On the other hand, the bandwidthrange of frequencies determines the differences between the highestfrequency signal component and the lowest frequency signal component.This is an electronic signal that is used on all transmissionmediums. Just like in the case of frequency of a signal, bandwidth ismeasured in hertz (cycles per second) (Sterbenz et. al., 2001).Bandwidth can be used in businesses to highlight the number ofstaffing available to accomplish a particular task (Iser, Minker &Schmidt, 2008).
This is a term used to measurethe number of units of information that a system can be able toprocess within a given period of time (Rodriguez-Ezpeleta, Hackenberg& Aransay, 2012). It is employed in varied processes ranging fromaspects of computer and network systems within an organization. Someof the ways through the productivity of such systems can be measuredinclude the response time and the speed taken by a particularapplication to be completed (ICIST (Conference), Dregvaite &Damaševičius, 2014). Additionally, the performance can bemeasured based on the period taken between one interactive userrequest and the subsequentreceipt of the response (Rodriguez-Ezpeleta, Hackenberg &Aransay, 2012).
Figure 2: Throughput
However, in most instances, theeffectiveness of a system has been determined by the measure ofcomparativeness of massive computers that run numerous programsconcurrently (Hill, Tiedeman & Wiley InterScience, 2007). In theearlier times, this was determined based on the number of batch jobscompleted on a daily basis. However, recent measures areincorporating a more sophisticated approach which focuses on specificoperations of the computer systems (Rodriguez-Ezpeleta, Hackenberg &Aransay, 2012). This has necessitated the introduction of units like“trillion floating-point operations per second” also known asTeraFLOPS and TFLOPS (Rodriguez-Ezpeleta, Hackenberg & Aransay,2012). These provide the metrics necessary for conducting acomparison of the cost of raw computing as desirable by themanufacturers. To achieve this, a benchmark is used to measure thethroughput. Regarding data transmission, network throughput is usedto gauge the amount of data that has been moved successfully from onelocation to another within a specified period (Rodriguez-Ezpeleta,Hackenberg & Aransay, 2012). As such, the measurements used arebits per second (bps), megabites per second (Mbps) and gigabites persecond (Gbps). With regards to storage systems, the throughput isused to refer to the amount of data that can be obtained and thenwritten to the storage medium (Rodriguez-Ezpeleta, Hackenberg &Aransay, 2012). Additionally, it reads from the media beforereturning to the requesting system which is measured in bytes persecond. The same case applies to the number of discrete input oroutput (I/O) operations that are responded to in a second (IOPS)(Rodriguez-Ezpeleta, Hackenberg & Aransay, 2012).
Latency refers to the delay aninput experiences in obtaining the desired outcome. However, ininformation technology, the term is used based on the context.Additionally, it will vary extensively depending on the system(Sterbenz et. al., 2001). This concept affects the usability andefficiency of both the electronic and mechanical devices (Hill,Tiedeman & Wiley InterScience, 2007). The performance ofcommunication devices will also massively be influenced by latency.The concept of latency in communication is highlighted in the processwhere live transmissions from various points hop between a groundtransmitter and satellite. As such, individuals that are connectingto these events from further distances have to wait for the responses(Holzbecher, 2012).
Figure 3: Latency
Latency in the context ofnetworks refers to the amount of time taken by a packet of data fromone designated location to another. Additionally, there are variouscontributors to network latency. One such donor is known aspropagation which refers to the time taken by a packet of data tomove from one location to the other while travelling at the speed oflight. Transmission depends on the medium being used (AmericanCongress on Surveying and Mapping, 1981). This can be in the form ofoptical fibres or wireless and other types. The size of the packetwill, therefore, determine the speed taken for the transmission.Relatively bigger packages take a relatively longer time to betransmitted to the intended location (Hill, Tiedeman & WileyInterScience, 2007).
Another contributor is the routerand other processing. This is I regards to the time taken by thegateways nodes to examine the header of the packets. At times, thenodes can change the heading as in the case of changing the hop countin the time-to-live field. The last contributors are the existence ofcomputer and storage delays. This is experienced within the networksand in instances where the hard disk is subjected to delays atintermediate devices that may include the switches and bridges.
This is a special case of networklatency since the web is an enormous wide-area network (WAN). Thelatency on the web is determined by various factors such as the sizeof the data packet and the medium. However, for the smaller networks,these are considered to be greater the number of hops overequipment, the transmission medium and the number of servers. Themeasurement of internet latency, therefore, starts at the exit of anetwork and the end of the requested data obtained from a webresource.
This is one of the most importantconcepts of determining internet latency. This is especially the casefor a WAN that is involved in directing other traffic and as suchexperiences delay as a result. The delay will be irrespective ofwhether a resource is being requested from a server on the LAN oranother computer as well as other areas of the internet. In a similarfashion, LAN users will experience delays when WAN is busy (AmericanCongress on Surveying and Mapping, 1981). The nature of the delayswould be felt even if the rest of hops including servers wereentirely free of any form of congestion.
This refers to the delays insounds being created and that which is heard. In the physical sense,the delay in sound is determined by the speed of sound which isdetermined by the medium through which the sound travels through. Asis common knowledge, the rate of sound is faster in the densermedium. In this case, the speed is highest in solids, relativelyslower in liquids and slowest in air. For this study, the speed ofsound used is assumed to be measured in dry air at room temperature,796 miles per hour (American Congress on Surveying and Mapping,1981). Concerning electronics, latency refers to the delayexperienced between the period of audio input and audio output. Thetime taken by such delays will massively depend on the hardware andsoftware used. The latter may include the operating system anddrivers that are utilized in computer audio. Individuals are likelyto notice latencies of 30 milliseconds regarding separate productionand the arrival of sound to the ear (American Congress on Surveyingand Mapping, 1981).
This refers to the total timetaken by an operation when the tasks are completed in linearworkflows. However, when analyzing parallel workflows, the latencywill be determined by the process with the slowest speed. The mosttime-consuming operation performed by a single task worker will,therefore, determine the operational latency in the parallelworkflows (Olukotun, Hammond & Laudon, 2007).
This refers to the delayexperienced concerning the input in a mechanical device or system tothe desired output. The delay in mechanical latency is determined byNewtonian physics-based limits of the system (Olukotun, Hammond &Laudon, 2007). However, there is an exception to quantum mechanics.An example of this type of delay would occur in the time taken toshift gear from the time the shift lever of a gear box or bicycleshifter was actuated (Olukotun, Hammond & Laudon, 2007).
Computerand Operating System Latency
The computer and operating systemlatency refer to a combination of the delay between an input/ commandand the desired output (Olukotun, Hammond & Laudon, 2007). Whendiscussing computer systems, latency is a term used to highlight allforms of delays and waiting that causes real or perceived responsetime beyond the desired levels. Some of the special contributors tothe delays in computer systems include mismatches that exist in dataspeeds between the microprocessors and the input and output devicesand inadequate data buffers. Additionally, other factors thatcontribute to the delays include performances of the hardware and thedrivers involved (Olukotun, Hammond & Laudon, 2007).
From the users’ perspective, onthe other hand, latency factors are discussed concerning the lagbetween an action and a response to it. An example, in this case,includes the 3D VR simulation whereby in using a helmet that providesa stereoscopic vision as well as head tracking, latency will bereferred to as the time between the moment of the computer detectshead motion and the moment the movement is displayed in the images.For multiplayer networks and other internet gaming, lower latencywill provide the best results as it facilitates the best game play aswell as enjoyability. In cases of significant lag, the controls wouldbe awkward since the player would be lagging behind the real-timeactions in the game as a result of the delays experienced concerninginformation getting to the computer (Gonzalez & Raman, 2015).
The challenges inherent inlatency will be experienced by individuals at different levels. Thisis likely to increase user annoyance while also having a significantimpact on the productive levels as the level increases above 30ms.However, the severity of the effects of such challenges is likely todepend on the nature of applications as well as the mitigatingtactics. For gaming activities, a latency of 90ms is still likely tofacilitate enjoyment. However, communication can be affected by thehardware problem, heavy traffic and configuration issues (Olukotun,Hammond & Laudon, 2007).
Testing of latency depends onfrom one application tot eh other. In several instances, the processwill involve the massive focus on the compound computer applicationsas well as the specialized knowledge of equipment used in theprocess. There is, therefore, need to be more conversant with variousprograms and applications. In other cases, the latency can bemeasured by using a stopwatch. About communication, the estimatedlatency of servers and equipment is established through running aping command. In addition to this, the trace route command is used togather information about latency. In other instances, the high-speedcameras can be utilized in determining the minute differences in theresponse times posted for the inputs to various other mechanical andelectronic systems (Oodan, 2003).
The process of cutting latencyinvolves a series of techniques. Such methods may include tuning,tweaking, and upgrading both computer hardware and software andmechanical systems. On the other hand, other techniques such asprefetching are used to remove latency within a computer. The abovemethod involves anticipating the need for data input requests.Multithreading and the use of parallelograms across numerousexecution threats are also another way through which a computersystem is ridden of latency. Other steps that can be taken in thequest to reduce latency in the systems include the increasing ofperformances through the unisntallation of unnecessary programs,upgrading or overlocking hardware, and the optimization of networkingand software configurations.
This is a term used to describeany form of deviation or the displacement of signal pulses that existin the high-frequency digital signal (McNeill & Rickets, 2009).The difference can be caused by various reasons and can be indifferent forms such as phase timing or the width of the signal pulseand the amplitude. Some of the causes of the jitters includeelectromagnetic interference ad crosstalk with other signals(Takasaki, 1991). Some of the ways through which the effects ofjitter can be analyzed include instances where display monitors arecaused to flicker and cause the loss of transmitted data betweenvarious networks (McNeill & Rickets, 2009). Additionally, jittercan be responsible for affecting the ability of processors incomputers as well as servers to perform their intended actions whilealso playing a massive role in the introduction of clicks and otherunwanted effects in the audio signals. The level and amount of jitterthat can be allowed within a system will depend on the particularapplication (Takasaki, 1991).
Figure 4: Jitter
With regards to the IP address,Jitter refers to the variations that are likely to occur in thelatency on a packet flow between two systems. This is usually evidentin instances whereby some packages takes relatively longer to travelfrom one system to the other. Some of the causes of jitter, in thiscase, include timing drift, route changes, and congestion of thenetworks (Takasaki, 1991).
The problems emanating fromjitter are more prominent in the real-time communication systems thatmay include IP telephony as well as video conferencing. Suchchallenges are also witnessed in the virtual desktop infrastructure(VDI) and hosted desktops. Jitter can, therefore, result inunintended deviation and inconsistencies in audio and video artifactsthat can subsequently downgrade the quality of communication (McNeill& Rickets, 2009).
However, in mitigating theeffects of jitter, jitter-buffers play a massive role. This can beaccomplished in the form of a network on a router or a switch as wellas a computer. In this case, the applications that consume suchnetwork packets will receive them from the buffer rather than obtainthem directly from their sources. These are fed to the buffer atregular intervals where they undergo smoothing out of the variationsthrough the effective timing of the packets flowing into the buffer(Oodan, 2003). There are additional techniques that can be utilizedin ensuring that jitter where multiple pathways for traffic areavailable (Takasaki, 1991). In this case, they involve the selectiverouting of vehicles on the most stable paths. Additionally, anothertechnique involves the selection of the route that is in the closestproximity to the packet delivery rate. The last two methods arelikely to achieve exemplary results when utilized in instances inwhich there are multiple pathways for traffic (WCNA & Zeng,2016).
In computing networks, packetdrop attack otherwise known as Blackhole attack refers to amalfunction whereby the router, which has the primary role ofrelaying packets, discards them instead (Yang, 2009). This form ofoccurrence is regarded as an example of the denial-of-service attack.Such an attack is caused by a router that has been compromised due tosome reasons. An example of a packet drop attack is through thedenial-of-service attack on the router referred to as DDoS (Yang,2009). The packet drop attack is very complicated to detect andprevent because packets are routinely dropped from a lossy network.In most instances the malicious router is likely to accomplish theattack selectively. This is evident in the fact that such a routermight drop packages for a specific network destination at aparticular time of the day. Such a scenario is referred to as a grayhole attack (Yang, 2009).
However, such an attack can bediscovered at a relatively faster rate in case the router attempts todrop all packets hat come in. This is accomplished through the use ofstandard networking tools such as trace route. Additionally, themoment the other routers realize that one of the routers that havebeen compromised all the traffic, the functioning ones embark on aprocess of removing the malfunctioning from their forwarding tables.In the end, no traffic will flow to the attack. A major problemarises when the malicious router drops packets for a particularperiod. In this case, a major issue will be experienced as some ofthe traffic will continue to flow across the network.
Figure 5: Packet Drops
The packet drop is used in mostinstances to attack wireless ad hoc networks. This is achieved basedon the fact that the wireless networks have a substantially differentarchitecture in comparison to the typical wired networks. Thisensures that a host is in a position to broadcast that it has theshortest path to a particular path. In such instances, all trafficwill then be directed to the host that has been compromised and thehost is therefore in a position to drop packets at will.Additionally, hosts are usually vulnerable to collaborative attacks.Such is the case where there is likelihood that the multiple hostswill become compromised thereby deceiving other hosts on the network.Such an occurrence relates to a mobile ad hoc network.
Packet loss results from thefailure of one or more transmitted packets to arrive at their finaldestination as in the case of digital communication. Packet lossoccurs in a variety of ways. It can lead to the loss of data as wellas being responsible for jitter in videoconferencing environment(Oodan, 2003). Additionally, it can also result in jitter andfrequency gaps. This is especially prevalent in transparent audiocommunication such as VoIP. However, its worst effects are witnessedwhereby packet loss causes distortion of the data received (Williams& Williams, 2004).
Bit error rate is a term thatfeatures significantly in the analysis of digital transmission(Shalkhauser, Budinger & Lewis Research Center, 1989). The numberof bit errors is therefore used to depict the instances where bits ofthe data stream of a communication channel that have been receivedare altered as a result of various factors such as distortion andnoise. Other factors that are likely to cause the alteration of thenumber of received bits of data stream include interference and bitsynchronization errors (Kartalopoulos, 2004). On the other hand, biterror rate (BER) defines the number of bit errors per unit time. Thebit error ratio, on the contrary, refers to the number of bit errorsover the number of bits that are transferred during a particular timeinterval. The ratio is a unitless performance measure since it isexpressed as a percentage (Gonzalez & Raman, 2015).
Figure 6: Error Rate
The bit error probability, on theother hand, refers to the expectation value of the bit error ratio.Packet error rate (PER) relates to the number of data packets thathave received incorrectly over the total number of packets receivedover a given period (Shalkhauser, Budinger & Lewis ResearchCenter, 1989). For any packet of data to be considered as incorrect,it has to exhibit at least a single bit that is erroneous in nature(Williams & Williams, 2004).
Various factors are likely toaffect the bit error rate. These factors include channel noise,distortion, bit synchronization problems, and interference. Othersinclude wireless multipath fading and attenuation (Shalkhauser,Budinger & Lewis Research Center, 1989). The error rate may,therefore, be improved through the selection of a high signalstrength and slow and robust modulation scheme (Kartalopoulos, 2004).The former cannot be achieved unless action causes cross-talk andmore bit errors. Another means through which the bit error rate maybe improved through the use of line coding scheme and through theapplication of channel coding schemes that include redundant forwarderror correction codes (Shalkhauser, Budinger & Lewis ResearchCenter, 1989).
The evaluation of bit error rateis accomplished through the use of stochastic computer simulations.However, the BER can be calculated analytically in the case of asimple transmission channel model, and source data model are assumed.The Bernoulli source is one such data source model. Various simplechannel models are used in information theory. The binary symmetricmodel is utilized in the analysis of decoding error such as in thecase of non-bursty bit errors in the transmission model. On the otherhand, additive white Gaussian noise (AWGN) channel is also used as asingle channel in the information theory. However, a worst casescenario is experienced in instances in an entirely random channelwhereby noise dominates the useful signal. The results are usually atransmission BER of 50%. However, for these results to be achieved,there should are assumptions that are made. The assumptions revolvearound the Bernoulli Binary data source and a binary symmetricalchannel (Kartalopoulos, 2004).
Bit error rate test (BERT) refersto a testing technique for digital communication circuits. It usespredetermined stress patterns that are composed of a sequence oflogical ones as well as zeros that have been generated by the testpattern generator. There are numerous types of bit error test stresspatterns as discussed below.
Pseudorandom binarysequence (PRBS): thisrefers to a pseudorandom binary sequence of N Bits. In this case, thepattern is utilized in measuring the eye mask in the TX-Data as wellas in the measure of jitter in both electrical and optical data links(Kartalopoulos, 2004).
Quasi-random signal source(QRSS): this is apseudorandom binary sequence tasked with the generation of everycombination of 10-word bits. To accomplish this, it repeats every1048575 words thereby suppressing the occurrence of consecutive zerosto a maximum of 14. This stress pattern contains both and high andlow-density sequences as well as those that can change from high tolow. The model is also used as the standard measurement of jitter(Shalkhauser, Budinger & Lewis Research Center, 1989).
3 in 24:this model contains the highest number of consecutive zeros with thelowest densities of 15 and 12.5% respectively. It operates bystressing the minimum frequencies while maximizing the number ofconsecutive zeros. H0owever, depending on the alignment of one bit toa frame, a D4 yellow alarm may be caused by the D4 frame format of 3in 24 (Kartalopoulos, 2004).
Bridge tap:the existence of bridge taps within a span can be detected by the useof various test patterns with a series of varying densities of onesand zeros. This kind of analysis provides 21 test patterns withinduration of 15 minutes. One or more bridge taps may be experienced asa result of the existence of signal errors. However, this model isonly regarded as sufficient for T1 spans that are used to transmitthe signal raw (Williams & Williams, 2004).
TheBit Error Rate Tester
This is electronic equipment thatis utilized in the measurement of quality of signals transmission.The methods employed signal transmission is a single component or acomplete system. There are a series of the main building blocks ofthe bit error rate tester, and they include the pattern generatorthat has a primary role of transmitting a particular test pattern tothe test system, and an error detector that has been connected to thetest system. The clock signal generator is used in thesynchronization of the pattern generators as well as the errordetector whereas the digital communication analyzer is optional inits display of the received signal. Finally, electrical-opticalconverter and the optical-electrical converter are used forconducting tests of the optical communication signals (Gonzalez &Raman, 2015).
In terms of computers, quality ofservice refers to the performance of the computer networks as seen bythe users. The quantification of the quality of service is a processthat requires the integration of numerous aspects that include errorrates, throughput, bit rates, and jitter. This concept of computer isimportant when analyzing the flow of traffic in networks. Therefore,quality of service is regarded as the ability of a network to provideusers with a certain degree of performance. It can be affected byboth human and technical factors. The former include delays,availability of services, and stability of service. On the otherhand, technical factors include effectiveness, maintainability,scalability, and grade of service.
Methodology Fulldetails of the Artifact
WAN, which stands for Wide AreaNetwork, is a computer network that covers a large geographical area,for example, a country. Birmingham and London`s subnets are shownbelow in detail. They are parts of the WAN model designed in theriverbed that I focused on during the simulation.
The Birmingham subnet consists ofthe Birmingham office and Birmingham router that is connected toother routers in the network that connect other subnets. There areswitches in the network they can be seen clearly in the Londonsubnet as shown in the figure below
The London subnet consists of theLondon office, the London router and the London switch. The Londonswitch controls web browsing server, VoIP server, FTP server, videoconferencing server, email server and Database server. The router isa networking device which forwards data between computer networks.Each subnet is connected to a router in the designed WAN networkmodel.
TheWhole WAN Model
I designed the WAN network modelusing the riverbed modeler which can assess the effects onperformance when subjected to various stresses of application thatwill be analyzed. The model simulation was conducted on differentapplications which include DB query, Email, FTP, HTTP, videoconferencing, voice delay and voice jitter. This was by theestablished computer networks standards which were applied onapplications of protocol standards, voice and data, distance,technologies and the number nodes.
Some of the applications improvedits quality of service while others reduced the quality of duty. Theresults produced will be able to show the optimum operation point ofperformance determined by the applications (WCNA & Zeng, 2016).The results were different as the applications work differently andaffect various aspects of the network.
The physical communicationnetwork is specified by the system editor (Xinjie Chang, 1999). Theposition and interconnections of the communicating tools are defined,For instance, whether a node is fixed or mobile. Customization of thelink is also possible depending on the simulation of thecommunication channel.
Running simulation comes afterthe designing of the model. Simulation editor is used to simulatingthe network created which in this case the specific informationneeded is extracted from the simulation depending on applicationspecific, statistics and behavioral characteristics (Xinjie Chang,1999).
Simulation gives the outputs indata generation. Different probes are used for various types of data.The analysis tool is then used to display the information ingraphical form.
FTP, File Transfer Protocol, isused for the transfer of computer files between a client and serveron a computer network. It runs over TCP. Transfer on the first linkvaries between the range 0.12 and 0.14. The second link is not insuch a good condition as FTP performance drops from around 0.17 toalmost 0.08 where it becomes constant but does not improve inperformance after around 4min.
FTP is one of the most common TCPbulk data transfers that enterprises use. However, it’s performanceis reduced dramatically across WAN, this can be seen in the graphabove. Alleviation of this challenge is usually unsuccessful even byadding WAN bandwidth as it does not handle the problem directly.Mitigation of the impact of network latency and intelligentlyreducing data size during transfer are essential for accelerating theFTP transfers over the WAN.
Transfer time is reduced, and thebandwidth usage is also reduced when using FTP compared to othertransfer protocols. It is faster. It is a client server that relieson two communication channels between client and server. Clientsrequest file download from the server. FTP enables a customer toupload, download, delete, rename, move and copy files on a server.
The quality of service of the WANis affected by the FTP over time it reduces. The effects areprominent when there’s congestion. The network will be slow andthus poor services to the clients. The quality of duty is good whenthe FTP carries fewer data, this guarantees faster speeds. In thiscase, the optimum operation point was at around 0.17 and 0.13 ataround 2min.
Voice delay is the timedifference between when a voice packet is sent and when it isreceived. It usually occurs due to network performance and thedistance between two nodes. Voice delay in both links reducesconsistently at around 2min. The first link which for WANApplications test-High speed links-DES-1 appears to be cutting frombelow 0.000060 to 0.000005 where it becomes constant and trails offat around 0.000000, slightly higher than zero. The optimum operationpoint of a WAN network, when subjected to voice delay, is at0.000000. At this point, the effects are negligible and thus cannotaffect the WAN network.
An increasing number of clientslead to more voice delay which causes a reduction in sound quality.It is important as voice information needs to be received in realtime. Conversation becomes difficult when the delay is too much.Sound delay reduces the quality of service.
Jitter is the deviation from trueperiodicity of a presumed periodic signal variance of time betweeneach packet arrival. This occurs due to network congestion. For thetest-high speed links, jitter is 0.000000 for about 9min. It thendrops slightly to below 0.000000 and then becomes constant. For thesecond client, WAN Applications test-low speed link1, jitterincreases spontaneously around the 2min to 4 minutes then it becomesconstant at around zero jitter, 0.000000. The increase in jitter isslight meaning it is still in good level.
Quality of service is measuredaccording to the Call quality, which rely heavily on the amount ofjitter. High voice jitter leads to poor call quality as voiceinformation will not be received in a timely fashion and thus theinformation will not be viable.
Optimum point of operation is atzero jitter. At this point, the WAN network is not affected by it, asit is absent. Jitter increases with increase in clients and links ona network. If it’s not corrected, the network will function poorly.One solution for jitter is the use of jitter buffers. Effects ofjitter can be corrected, VoIP endpoints collect packets in a bufferand they are put back together in the proper timing and order beforethey are heard by the receiver. This is a balancing act, bufferprocessing adds delay time to the call, this means, the bigger thebuffer, the longer the delay will be. If the voice packet arriveswhen the buffer is full, they are discarded.
DB query is a databaseapplication. Its response time was about two minutes. Database queryrequests for information from a database. A DB query can either be aselect query or an action query. A select query is a data retrievalquery while action query can ask for additional operations on thedata for example deletion, updating, and insertion.
Subjecting the WAN network to theDB query produced the above graph. We monitored the DB querysimulation for ten minutes. For the WAN applications test-high-speed links- DES- 1, it started off at point 0 until about 2minwhere it went up to 2 and then dropped back to 0. It kept going up to2 and back to 0 until about the fifth minute. At some points, timetaken for going up and down was shorter than others it wasirregular. At around the 6min, it shot up to 3, the highest pointreached. After 5min minute, it didn’t go beyond 1 and always cameback to 0 up to the 10th minute. The optimum point of operation was3.
The WAN application test-lowspeed link1-DES-1 behaved the same way as the first link but withdifferent readings and different points of optimum operation. It wentup to 1 and came down to 0 up to the forth minute. At 4min, it wentup to 2, dropped to 0. At around the 5th minute, it went up to 3,this was the highest point reached. It came down to 0, went up to 1,dropped to zero again and maintained its position at 0 until around7min where it went on with 1 being its peak until about the 8thminute. It went up to 2 at around the 8th, 9th and 10th minutes. Theoptimum operation point was 3.
BD query has various effects onthe quality of service. Quality of service is reduced when there is alarge query to be processed. This slows down the network and it canbe momentarily unavailable. On the other hand, queries can have apositive effect on the quality of service. DB queries operate in highspeed thus large amounts of data can be retrieved from a databasequickly and efficiently. Queries save on time as the relevantinformation is extracted from a database. It also minimizes on datainconsistency which exists when a data has different versions indifferent places.
HTTP,Hypertext Transfer Protocol, is an application-level protocol fordistributed, collaborative, hypermedia information systems. It isused by the World Wide Web, www. It defines how messages areformatted, transmitted, and the actions were taken by web servers andbrowsers about commands. Hypertext being a structured text useslogical links between nodes containing texts. HTTP, therefore, is theprotocol to transfer hypertext simply put.
HTTP over theWAN network produced the graph shown above during simulation. Thefirst link started rising at about 3min from 0.29 to a point ofapproximately 0.64 at 4min. It dropped to 0.55 and then maintainedthat position up to the 7th minute where it had intermittent risewith breaks till a maximum point, 0.64, at around the 8th minute fromwhere there was a drop to 0.60 at 9m after which there is no effectwith change in time up to 10min. The second link, WAN applicationstest-Low speed links1-DES-1, the graph shows that it starts off from0.49, and there is no effect on it until 4min where it drops to 0.37.It then rises slightly to 0.41 and becomes constant up to 6min whereit drops to 0.36 and maintains. There is a steady growth with breaksup to 0.45 at 9min from where there’s no change with time up to the10th minute. The optimum operation point of performance is 0.64 forHTTP simulation.
The quality ofservice improves with HTTP on the network. HTTP reduces packet count.HTTP downloads each file over connections that are independent andthen closes them thus reducing interception risk during transmissionas the link does not go past the transfer of a single element on aweb page. Each item to be transferred requires a different amount ofresources to store and download HTTP enables multiple connections todownload separate elements concurrently thus speeds up thetransmission. The various drops in the graph may be caused by theneed for multiple connections to transmit a typical web page this isa drawback.
Stateless isone shortcoming of the HTTP. It being a stateless protocol is becauseeach command is executed independently without any knowledge ofprevious commands. Thus, implementation of websites that have anintelligent reaction to user input is hard.
VideoConferencing (overlaid statistics)
Videoconferencing is a live connection between people who are in separatelocations for the purpose of communication. The communication is bysimultaneous two-way video and audio transmissions. At its simplest,videoconferencing provides transmission of static images and textbetween two locations. At its most sophisticated, it transmitsfull-motion video images and audio of high quality between multiplelocations.
The WANnetwork was again subjected to videoconferencing, and the stress onit was assessed. The result was produced as shown in the abovegraph. At around 2min, the first link, WAN Applications test-Highspeed links-DES-1,dropped from approximately 0.000027 to around0.000006 in about 1 minute time. It then rose up to around 0.000031at the 4th minute. There was a slight gradual drop to 0.000026 ataround the 7th minute. There was a steady increase after the 8thminute to the 10th minute, around0.000032.
The secondlink, WAN Application test-low speed links1-DES-1, maintained itsposition at 0.000000 throughout the 10 minutes of the procedure.
VideoConferencing (stacked statistics)
Stackedstatistics of the video conferencing application effect on WANnetwork is as shown above. The results were different from the oneproduced by the overlaid one. The first graph shows average in videoconferencing packet delay variation of the first link, WANApplications test-High speed links-DES-1. It starts rising fromaround 0.000000050 to 0.000000060 at the 2nd minute from where itdrops steadily to 0.000000020 at the 3rd minute and then dropsfurther to around 0.000000010 at 7min. There’s a rise after 7min to0.000000020 at 10min.
The graph fromthe second link, WAN Application test-High speed links-DES-1, had thesame shape as the one of the first connections in the overlaidstatistics. There is a steady drop 0.000026 at around the 7th minutewhere it rises to around0.000031 at 10min.
The optimumoperation point of performance for the video conferencing, overlaidstatistics was at 0.000032. At this point, the WAN network’sperformance was optimum according to the simulation for the given 10minutes. Boardrooms that are fixed on a particular site use dedicatednetwork routes and internet access that are required to achievefull-screen full motion video. Optimum operation is completed whenall factors are considered and the nodes are set correctly. Forexample, video compression quality should be high and the bandwidthshould be adequate as video conferencing application consumesbandwidth similar to web browsing, web conferencing, desktop sharingand VoIP.
Quality ofservice varied. At some point, it was good and at some point it waspoor. Video conferencing affects the quality of service of WANnetwork. If the service providers’ standards are weak, then thequality of duty is poor but with good standards, quality of serviceis improved.
Videoconferencing is made possible by use of a video camera, microphone,and mounted speakers. The participants speak to one another, as theydo so, their voices are carried over the network to the other’sspeakers and the images appear in front of the video camera isdisplayed in a window on the other participants’ monitor.
Email which isthe short form of electronic mail refers to a system of exchangingdigital messages between computer users. The messages are sent viatelecommunication links over communication networks. Typically, emailmessages are notes entered from the keyboard or electronic files.There are some electronic mail systems that are confined to a singlecomputer system and others that have gateways to other computersystems that enables sending of emails anywhere in the world.
Emailapplication affects WAN network as shown in the graph above. Thegraph shows the average in Email download response time. WANApplications test-High speed links-DES-1, first link, maintained itsposition at around 0.014. It wasn’t affected by the change in time.The second link, WAN Applications test-High speed links1-DES-1,changed with change in time. It rose spontaneously from 0.017 to0.025 at around 2.5min. It maintained its position at 0.025 up to the7thminute where there was a slight rise to slightly above 0.025,approximately 0.0253 after which the position was maintained up tothe 10thminute, there was no effect with change in time.
Multipleprotocols within the TCP/IP suite are used by the Email. Theseinclude SMTP, POP, IMAP and HTML. SMTP is used to send messages. POPand IMAP protocols are used for retrieval of messages from the emailserver. HTML allows the email to be formatted. Original emailstandard supported plain text only, with time, it has evolved tosupport rich text with custom formatting. HTML email messages caninclude images, links and CSS layouts.
Based on the above analysis,various conclusions can be made regarding the use of the riverbedmodeller software. Thenetwork simulation conducted predicted the behavior of the WANnetwork model designed in riverbed modeler when subjected todifferent applications. Optimum points of operation varied with theapplications. Quality of service was affected negatively by someapplications while others affected it positively.
File TransferProtocol and voice delay reduced the networks performance greatly.This is due to reduction of transfer time and bandwidth usage.Quantity of service reduces when subjected to FTP. High voice delayled to poor network performance. The performance was optimum whenthere was no voice delay. Voice delay reduced the quality of serviceas communication needs to be in real-time. There was optimumperformance when voice delay was absent.
Voice jitterreduced the quality of service as there was variance of time betweenpacket arrivals. The designed WAN network functioned poorly whensubjected to voice jitter. However, at zero jitter the networkoperated normally.
DB queryeffect on the network was completely different from the stressessubjected to it by the other applications. It kept shifting from highto low spontaneously with no intermediates. It reduced the quality ofservice of the network when a large query had to be processed andimproved the QoS when no query was being processed optimum point.
HTTP improvedthe quality of service of the network. This is because it reduces thepacket count being transferred on the network thus improving itsperformance.
Videoconferencing and email effect on the WAN network varied with theamount of data transmission. There was no effect on the network’sperformance when data processing was low or absent. Negativeperformance occurred when data traffic was high due to congestionotherwise, their effect on quality of service was positive.
In conclusion,the project was successful. The model of WAN designed worked properlyas expected and viable results were produced.
WCNA(Conference), & Zeng, Q.-A. (2016). Wirelesscommunications, networking and applications: Proceedings of WCNA2014.
ICIST(Conference), Dregvaite, G., & Damaševičius, R.(2014). Informationand software technologies: 20th International Conference, ICIST 2014,Druskininkai, Lithuania, October 9-10, 2014. Proceedings.
Aboulia, E.(2011). NetworkSimulation Experiments Manual.Burlington: Elsevier Science.
Holzbecher,E. O. (2012). Environmentalmodeling: Using MATLAB®.
Hill, M.C., Tiedeman, C. R., & Wiley InterScience (Online service).(2007). Effectivegroundwater model calibration: With analysis of data, sensitivities,predictions, and uncertainty.Hoboken, N.J: Wiley-Interscience.
Gonzalez-Prida,V., & Raman, A. (2015). Promotingsustainable practices through energy engineering and assetmanagement.
Williams,B., & Williams, J. (2004). Tradingchaos: Maximize profits with proven technical techniques.New York: Wiley.
AmericanCongress on Surveying and Mapping. (1981). ACSMbulletin.Falls Church, Va.: American Congress on Surveying and Mapping.
Kartalopoulos,S. V. (2004). Opticalbit error rate: An estimation methodology.Piscataway, N.J: IEEE Press [u.a..
Shalkhauser,M. J. W., Budinger, J. M., & Lewis Research Center.(1989). Digitallymodulated bit error rate measurement system for microwave componentevaluation.Washington, D.C: National Aeronautics and Space Administration,Office of Management, Scientific and Technical Information Division.
Iser, B.,Minker, W., & Schmidt, G. (2008). Bandwidthextension of speech signals.New York: Springer.
Yang, H.(2009). Ahybrid packet loss recovery technique in wireless ad hoc networks.Ottawa: Library and Archives Canada = Bibliothèque et ArchivesCanada.
Oodan, A.P. (2003). Telecommunicationsquality of service management: From legacy to emerging services.London: Institution of Electrical Engineers.
Olukotun,O. A., Hammond, L. S., & Laudon, J. P. (2007). Chipmultiprocessor architecture: Techniques to improve throughput andlatency.San Rafael, Calif.: Morgan & Claypool Publishers.
Sterbenz,J. P. G., Touch, J. D., Escobar, J., Krishnan, R., Qiao, C., &Chapin, A. L. (2001). High-speednetworking: A systematic approach to high-bandwidth low-latencycommunication.New York: Wiley.
McNeill, J.A., & Ricketts, D. (2009). Thedesigner`s guide to jitter in ring oscillators.New York: Springer.
Takasaki,Y. (1991). Digitaltransmission design and jitter analysis.Boston: Artech House.
Rodríguez-Ezpeleta,N., Hackenberg, M., & Aransay, A. M. (2012). Bioinformaticsfor high throughput sequencing.New York, NY: Springer.
A:Birmingham SubnetB:London SubnetC:The Whole WAN ModelD:FTPE:Voice DelayF:Voice jitterG:DB QueryH:HTTPI:Video Conferencing (overlaid statistics)J:Video Conferencing (stacked statistics)K:Email