1 deployment of voip in data networks deployment of voip in data networks dr. khaled salah...

1

Deployment of VoIP in Deployment of VoIP in Data NetworksData Networks

Dr. Khaled SalahDr. Khaled Salah

[email protected]@kfupm.edu.sa

2

A step-by-step MethodologyA step-by-step Methodology

To deploy VoIP in Data NetworksTo deploy VoIP in Data Networks

K. Salah 3

OutlineOutline

Introduction and challenging Introduction and challenging questionsquestions

Existing toolsExisting toolsDrawbacks of existing toolsDrawbacks of existing toolsEight-step MethodologyEight-step MethodologyCase StudyCase StudyAnalytical toolAnalytical tool

K. Salah 4

IntroductionIntroduction

Importance of VoIPImportance of VoIP Unification of data and voice networks It is easier to run, manage, and maintain. Future – NGN and Triple Play

Existing IP networks are best effort and VoIP requires QoSExisting IP networks are best effort and VoIP requires QoS

Challenging questionsChallenging questions What are the QoS requirements for VoIP? How will the new VoIP load impact the QoS of currently running

network services and applications? Will my existing network support VoIP and satisfy the standardized

QoS requirements? If so, how many VoIP calls can the network support before upgrading

prematurely any part of the existing network hardware?

K. Salah 5

Existing ToolsExisting Tools

Ample of commercial toolsAmple of commercial tools NetIQ Brix Networks Agilent Cisco Avaya Siemens

Uses two common approaches for assessing the Uses two common approaches for assessing the deployment of VoIPdeployment of VoIP Take network measurements and then predict the

readiness based on the health of network Inject real VoIP traffic and measure QoS

K. Salah 6

Drawbacks of Existing ToolsDrawbacks of Existing Tools

CostCost

Injection approach can be intrusive to operation of Injection approach can be intrusive to operation of existing networkexisting network

None offers a comprehensive approach or methodology None offers a comprehensive approach or methodology for successful VoIP deployment. for successful VoIP deployment. No answers to all challenging questions, e.g.

Number of calls Call distribution Call flow Future growth Impact on existing network apps

K. Salah 7

Case StudyCase Study

Floor 1

Internet

7

Switch 1 Switch 2

. . .

User PCs

Workgroupserver

Floor 3

. . .

User PCs

Floor 2

. . .

User PCs

Databaseserver E-Mail

server

Fileserver HTTP

server

Web &cacheproxy

Firewall

Router

Printerserver

Workgroupserver

Printerserver

Workgroupserver

Printerserver

K. Salah 8

MethodologyMethodology

Determine VoIP characteristics and requirementsDetermine VoIP characteristics and requirements Determine VoIP traffic flow and call distributionDetermine VoIP traffic flow and call distribution Define performance thresholds and growth capacityDefine performance thresholds and growth capacity Perform network measurementsPerform network measurements Early modifications to existing network Early modifications to existing network Theoretical AnalysisTheoretical Analysis OPNET SimulationOPNET Simulation Comparison of Simulation and AnalysisComparison of Simulation and Analysis Final modifications to existing networkFinal modifications to existing network

K. Salah 9

Perform networkmeasurements

Determine VoIP Characteristics,requirements and assumptions

Define performance thresholdsand future growth capacity

Upfront network assessmentor modifications

Analysis Simulation

PilotDeployment

Start

End

(3)(1)

(5)

(6) (7)

(8)

Determine VoIP traffic flowand call distribution

(2) (4)

K. Salah 10

Perform network measurements

Determine VoIP Characteristics, requirements and assumptions

Define performance thresholds and future growth capacity

Upfront network assessment or modifications

Analysis Simulation

Pilot Deployment

Start

End

)3()1(

)5(

)6( )7(

)8(

Determine VoIP traffic flow and call distribution

)2( )4(

K. Salah 11

VoIP Traffic Characteristics, VoIP Traffic Characteristics, Requirements, and AssumptionsRequirements, and Assumptions

A point-to-point conversation for all VoIP calls with no A point-to-point conversation for all VoIP calls with no call conferencing call conferencing

HardwareHardware Gatekeeper or CallManager

handles signaling for establishing, terminating, and authorizing connections of all VoIP calls.

H.323 or SIP Gateway

responsible for converting VoIP calls to/from the Public Switched Telephone Network (PSTN).

VoIP Terminal IP phones Desktop with IP SoftPhones

As an engineering and design issue, the placement of these nodes in the network becomes crucial.

K. Salah 12

VoIP end-to-end ComponentsVoIP end-to-end Components

EncoderEncoder PacketizerPacketizer Playback BufferPlayback Buffer DecoderDecoder

Encoder Packetizer

Sender Receiver

NetworkPlayback

Buffer

Depacketizer Decoder

K. Salah 13

Common ITU-T codecs and their defaults Common ITU-T codecs and their defaults

G.711u gives a MOS of 4.4G.711u gives a MOS of 4.4 Other codes use (to decrease rate):Other codes use (to decrease rate):

compression silence suppression packet loss concealment encapsulating voice packets in one Ethernet frame

K. Salah 14

End-to-End Delay for a Single Voice PacketEnd-to-End Delay for a Single Voice Packet

The end-to-end delay is sometimes referred to by The end-to-end delay is sometimes referred to by M2EM2E or Mouth-to-Ear or Mouth-to-Ear delay delay

G.714 imposes a maximum total one-way packet delay of G.714 imposes a maximum total one-way packet delay of 150ms 150ms end-end-to-end for VoIP applications to-end for VoIP applications

200ms was found to be acceptable by experimentation200ms was found to be acceptable by experimentation Sources of delay:Sources of delay:

(i) encoding, compression, and packetization delay at the sender(ii) propagation, transmission and queuing delay in the network(iii) buffering, decompression, depacketization, decoding, and

playback delay at the receiver.

Encoder Packetizer

Sender Receiver

NetworkPlayback

Buffer


K. Salah 15

VoIP Traffic Characteristics and RequirementsVoIP Traffic Characteristics and Requirements

M2E delay for a single callM2E delay for a single call 150ms according to G.714 Sender: 25 ms Receiver: 45 ms

Higher than the sender. It includes jitter buffer delay which is at most 2 packets or 40 ms

Network: 80 ms

Encoder Packetizer

Sender Receiver

NetworkPlayback

Buffer


K. Salah 16

Bandwidth for a Single Call Bandwidth for a Single Call

The required bandwidth for a single call, one direction, is 64 kbps. The required bandwidth for a single call, one direction, is 64 kbps. G.711 codec samples 20ms of voice per packet. Therefore, 50 G.711 codec samples 20ms of voice per packet. Therefore, 50

such packets need to be transmitted per second. such packets need to be transmitted per second. Each packet contains 160 voice samples in order to give 8000 Each packet contains 160 voice samples in order to give 8000

samples per second. PCM sampling & quantization is done every samples per second. PCM sampling & quantization is done every 125us.125us.

Each packet is sent in one Ethernet frame. With every packet of Each packet is sent in one Ethernet frame. With every packet of size 160 bytes, headers of additional protocol layers are added. size 160 bytes, headers of additional protocol layers are added. These headers include RTP + UDP + IP + Ethernet with preamble These headers include RTP + UDP + IP + Ethernet with preamble of sizes 12 + 8 + 20 + 26, respectively. of sizes 12 + 8 + 20 + 26, respectively.

Therefore, a total of 226 bytes, or 1808 bits, needs to be Therefore, a total of 226 bytes, or 1808 bits, needs to be transmitted 50 times per second, or 90.4 kbps, transmitted 50 times per second, or 90.4 kbps, in one direction.in one direction.

For both directions,For both directions, the required bandwidth for a single call is 100 the required bandwidth for a single call is 100 pps or 180.8 kbps assuming a symmetric flow. pps or 180.8 kbps assuming a symmetric flow.

K. Salah 17

Other AssumptionsOther Assumptions

Voice calls are symmetric and no voice conferencing Voice calls are symmetric and no voice conferencing We also ignore the signaling traffic generated by the We also ignore the signaling traffic generated by the gatekeepergatekeeper. .

Worst-case scenario is considered signaling traffic involving the gatekeeper is mostly generated prior to

the establishment of the voice call and when the call is finished. This traffic is relatively small compared to the actual voice call traffic.

gatekeeper generates no or very limited signaling traffic throughout the duration of the VoIP call for an already established on-going call

No QoS mechanisms that can enhance the quality of No QoS mechanisms that can enhance the quality of packet delivery in IP networks, such aspacket delivery in IP networks, such as

IEEE 802.1p/Q IETF’s RSVP DiffServ MPLS

K. Salah 18





Analysis Simulation

Pilot Deployment

Start

End

)3()1(

)5(

)6( )7(

)8(


)2( )4(

K. Salah 19

VoIP Traffic Flow and Call DistributionVoIP Traffic Flow and Call Distribution Knowing the current telephone call usage or volume of the enterprise is an Knowing the current telephone call usage or volume of the enterprise is an

important step for a successful VoIP deployment. important step for a successful VoIP deployment. Collecting statistics about of the present call volume and profiles is Collecting statistics about of the present call volume and profiles is

essential. essential. Sources

PBX database Telephone records Billing

Key characteristics of existing calls can include the number of calls, number Key characteristics of existing calls can include the number of calls, number of concurrent calls, time, duration, etc of concurrent calls, time, duration, etc

We want to investigate if these characteristics can be still met when migrating to VoIP

Locations of the call endpoints, i.e., the sources and destinations, as well as Locations of the call endpoints, i.e., the sources and destinations, as well as their corresponding path or flow their corresponding path or flow

Call distribution must include percentage of calls within and outside of a Call distribution must include percentage of calls within and outside of a floor, building, department, or organization. floor, building, department, or organization.

As a good capacity planning measure, it is recommended to base the VoIP As a good capacity planning measure, it is recommended to base the VoIP call distribution on the busy hour traffic of phone calls for the busiest day of call distribution on the busy hour traffic of phone calls for the busiest day of a week or a month. a week or a month.

The projected extra calls need to be also combined with current statisticsThe projected extra calls need to be also combined with current statistics

K. Salah 20

Call DistributionCall Distribution

External

Total calls

1/3 2/3

F1 F2 F3

1/31/31/3

F1 F2 F3

1/31/31/3

F1 F2 F1 F3 F2 F3

Intra-floor Inter-floor

1/32/3

Internal

1/31/31/3

K. Salah 21





Analysis Simulation

Pilot Deployment

Start

End

)3()1(

)5(

)6( )7(

)8(


)2( )4(

K. Salah 22

Define Performance Thresholds and Define Performance Thresholds and Growth CapacityGrowth Capacity

The maximum tolerable end-to-end delayThe maximum tolerable end-to-end delay determined by the most sensitive application to run on the

network 150ms for VoIP

It is imperative to note that if the network has certain delay-sensitive applications, the delay for these applications should be monitored, when introducing VoIP traffic, such that they do not exceed their required maximum values.

The utilization bounds or thresholds of network resourcesThe utilization bounds or thresholds of network resources Factors to consider: current utilization, future plans, and foreseen

growth of the network. It is extremely important not to utilize fully the network resources.

Packet lossPacket loss Depends on network service or application For VoIP, 0.1% to 5% packet loss is acceptable

K. Salah 23

Future GrowthFuture Growth

What is the projected growth in users, What is the projected growth in users, network services, business, etc.?network services, business, etc.?

In our study we will ascertain that 25% of In our study we will ascertain that 25% of the available network capacity is reserved the available network capacity is reserved for future growth and expansion. for future growth and expansion. we will apply this evenly to all network

resources of the router, switches, and switched-Ethernet links.

K. Salah 24





Analysis Simulation

Pilot Deployment

Start

End

)3()1(

)5(

)6( )7(

)8(


)2( )4(

K. Salah 25

Perform Network measurementsPerform Network measurements

Need to characterize the existing network Need to characterize the existing network traffic load, utilization, and flow traffic load, utilization, and flow

Background traffic profilingBackground traffic profilingAvailable tools:Available tools:

Open-source MRTG, STG, SNMPUtil, and GetIF

Commercial HP OpenView, Cisco Netflow, Lucent VitalSuite,

Patrol DashBoard, Omegon NetAlly, Avaya ExamiNet, NetIQ Vivinet Assessor, etc.

K. Salah 26

Perform Network measurementsPerform Network measurements

Network measurements must be performed for network Network measurements must be performed for network elements such as routers, switches, and links. elements such as routers, switches, and links.

Numerous types of measurements and statistics can be Numerous types of measurements and statistics can be obtained using measurement tools. obtained using measurement tools. As a minimum, traffic rates in bps (bits per second) and pps

(packets per second) must be measured for links directly connected to routers and switches.

To get adequate assessment, network measurements To get adequate assessment, network measurements have to be taken over a long period of time, at least 24-have to be taken over a long period of time, at least 24-hour period. hour period. Sometimes it is desirable to take measurements over several

days or a week.

K. Salah 27

K. Salah 28

Worst-case network measurementsWorst-case network measurements

K. Salah 29





Analysis Simulation

Pilot Deployment

Start

End

)3()1(

)5(

)6( )7(

)8(


)2( )4(

K. Salah 30

Upfront Network Assessment and Upfront Network Assessment and ModificationsModifications

Examine if any immediate modifications are Examine if any immediate modifications are necessarynecessary may include adding and placing new servers or devices,

upgrading PCs, and re-dimensioning heavily utilized links.

As a good upgrade rule, topology changes need As a good upgrade rule, topology changes need to be kept to minimum and should not be made to be kept to minimum and should not be made unless it is necessary and justifiable. Over-unless it is necessary and justifiable. Over-engineering the network and premature engineering the network and premature upgrades are costly and considered as poor upgrades are costly and considered as poor design practicesdesign practices

K. Salah 31

Changes to topologyChanges to topology

Links are underutilized, no need for 1G linksLinks are underutilized, no need for 1G links Shared links must be replaced with full-duplex switchedShared links must be replaced with full-duplex switched

shared Ethernet offers zero QoS and are not recommended for real-time and delay-sensitive applications as it introduces excessive and variable latency under heavy loads and when subjected to intense bursty traffic

Add Add gatekeepergatekeeper and and gatewaygateway connecting the gatekeeper to Switch 1 is practical in order to

keep the traffic local. Connecting the gateway to Switch 2 balances the projected load

on both switches. It is more reliable and fault-tolerant not to connect both nodes to

the same switch in order to eliminate problems that stem from a single point of failure.

K. Salah 32

Original TopologyOriginal Topology

Floor 1

Internet

7

Switch 1 Switch 2

. . .

User PCs

Workgroupserver

Floor 3

. . .

User PCs

Floor 2

. . .

User PCs

Databaseserver E-Mail

server

Fileserver HTTP

server

Web &cacheproxy

Firewall

Router

Printerserver

Workgroupserver

Printerserver

Workgroupserver

Printerserver

K. Salah 33

New Topology with VoIP ComponentsNew Topology with VoIP Components

Floor 1

Internet

7

Switch 1 Switch 2

Floor 3Database

server E-Mailserver

Fileserver HTTP

server

Web &cacheproxy

Firewall

Router

Gatekeeper

PSTN

VoIPGateway

. . .

Workgroupserver

Printerserver

MultimediaPCs

Floor 2

. . .

Workgroupserver

MultimediaPCs

Printerserver

. . .

Workgroupserver

MultimediaPCs

Printerserver

K. Salah 34





Analysis Simulation

Pilot Deployment

Start

End

)3()1(

)5(

)6( )7(

)8(


)2( )4(

K. Salah 35

The analytical approachThe analytical approach

Bandwidth bottleneck analysisBandwidth bottleneck analysis

Delay analysisDelay analysis

The actual number of VoIP calls that the network can sustain and The actual number of VoIP calls that the network can sustain and support is bounded by those two metrics. support is bounded by those two metrics.

Depending on the network under study, either the available Depending on the network under study, either the available bandwidth or delay can be the key dominant factor in determining bandwidth or delay can be the key dominant factor in determining the number of calls that can be supported.the number of calls that can be supported.

K. Salah 36

BW bottleneck analysisBW bottleneck analysis

K. Salah 37

Network Delay AnalysisNetwork Delay Analysis

Poisson VoIP TrafficPoisson VoIP Traffic Using Jackson TheoremUsing Jackson Theorem Links: M/D/1Links: M/D/1 Router and Switches: M/M/1Router and Switches: M/M/1

K. Salah 38

K. Salah 39

Network Capacity AlgorithmNetwork Capacity Algorithm

i.i. Add background trafficAdd background traffic

ii.ii. Add one call based on distribution and flowAdd one call based on distribution and flow

iii.iii. For each node calculate the new arrival rate – For each node calculate the new arrival rate – not all nodes are affected.not all nodes are affected.

iv.iv. Compute packet network delay for all flows by Compute packet network delay for all flows by summing up individual delays per nodesumming up individual delays per node

v.v. If network delay < 80 ms, go to If network delay < 80 ms, go to iiii, otherwise , otherwise STOP. STOP.

K. Salah 40

K. Salah 41

Worst incurred delay vs. number of VoIP callsWorst incurred delay vs. number of VoIP calls

K. Salah 42

Analytical ToolAnalytical Tool

Generic Generic

GUIGUI With drag-and-drop features

Analytical engineAnalytical engine BW bottleneck analysis Compute iteratively the network delay

K. Salah 43





Analysis Simulation

Pilot Deployment

Start

End

)3()1(

)5(

)6( )7(

)8(


)2( )4(

K. Salah 44

SimulationSimulation

Using Using OPNETOPNET

Will be discussed in great detail in next presentation

K. Salah 45

ComparisonComparison

A way to validate results of both simulation A way to validate results of both simulation and analysis (or expert intuition).and analysis (or expert intuition).

K. Salah 46





Analysis Simulation

Pilot Deployment

Start

End

)3()1(

)5(

)6( )7(

)8(


)2( )4(

K. Salah 47

Pilot DeploymentPilot Deployment

Before embarking on changing any of the network equipment, it is Before embarking on changing any of the network equipment, it is always recommended to build a pilot deployment of VoIP in a test always recommended to build a pilot deployment of VoIP in a test lab to ensure smooth upgrade and transition with minimum lab to ensure smooth upgrade and transition with minimum disruption of network services. disruption of network services.

A pilot deployment is the place for the network engineers, support A pilot deployment is the place for the network engineers, support and maintenance team to get firsthand experience with VoIP and maintenance team to get firsthand experience with VoIP systems and their behavior. systems and their behavior.

New VoIP devices and equipment are evaluated, configured, tuned, New VoIP devices and equipment are evaluated, configured, tuned, tested, managed, monitored, etc.tested, managed, monitored, etc.

Get comfortable with how VoIP works, how it mixes with other traffic, Get comfortable with how VoIP works, how it mixes with other traffic, how to diagnose and troubleshoot potential problems. how to diagnose and troubleshoot potential problems.

Simple VoIP calls can be set up and some benchmark testing can Simple VoIP calls can be set up and some benchmark testing can be performed. be performed.

K. Salah 48

To SummarizeTo Summarize

A step-by-step methodology on how VoIP can be A step-by-step methodology on how VoIP can be deployed successfullydeployed successfully

The methodology can help network researchers and The methodology can help network researchers and designers to determine quickly and easily how well VoIP designers to determine quickly and easily how well VoIP will perform on a network prior to deployment. will perform on a network prior to deployment.

Prior to the purchase and deployment of VoIP Prior to the purchase and deployment of VoIP equipment, it is possible to predict the number of VoIP equipment, it is possible to predict the number of VoIP calls that can be sustained by the network while calls that can be sustained by the network while satisfying QoS requirements of all existing and new satisfying QoS requirements of all existing and new network services and leaving enough capacity for future network services and leaving enough capacity for future growth. growth.

K. Salah 49

Perform networkmeasurements

Determine VoIP Characteristics,requirements and assumptions

Define performance thresholdsand future growth capacity

Upfront network assessmentor modifications

Analysis Simulation

PilotDeployment

Start

End

(3)(1)

(5)

(6) (7)

(8)

Determine VoIP traffic flowand call distribution

(2) (4)

1 deployment of voip in data networks deployment of voip in data networks dr. khaled salah...

Documents