Tdc 464 Converged Multimedia Networks - Lecture 8

Front 1

When we deploy VoIP there are two important factors

Back 1

Voice Quality of MOS 4 or better.

No blocking, meaning call will go through

Front 2

Is there a need for QoS on PSTN?

Back 2

No. PSTN uses circuit switched network and dedicated resources.

VoIP uses packet-switch network where there is contention of multiple stream

Contention of voice and data services on the same physical facility.

Front 3

What are the drivers for QoS for VoIP?

Back 3

Customers are willing to pay more for QoS service

Front 4

Generally speaking IP is a best effort service. What does best effort mean.

Back 4

There is no priority.

Front 5

The need for QoS

Back 5

A collective measure of the level of services
For a particular application
Performance criteria
Availability, throughput, connection setup time, percentage of successful transmissions, time to fault detection and correction
Bandwidth, packet loss, delay and jitter
IP is a best-effort service
Well suited to non-real-time data communication
To attract and retain paying subscribers
Circuit switching has a distinct advantage
But ill-suited to other forms of communication
IP network: solutions for the QoS are needed
Resource-reservation techniques

Front 6

Why is MPLS a popular protocol for Service Provider

Back 6

Because MPLS can allow Service Provider to guarantee service

Front 7

TCP and UDP

Back 7

TCP
Error-free, in-sequence delivery
But relatively long delay due to acknowledgement and congestion control
UDP
Fine for transporting voice
Provided that
Low packet loss
No congestion on the network. If there is congestion, some packets are lost.
Traffic in the network can be bursty and unpredictable
If there is significant packet loss, a speaker may be forced to repeat what he/she just said

Front 8

End-to-End QoS

Back 8

QoS must be end-to-end
The support of all network components in the chain
Service Level Agreement (SLA)
Differentiated services for paying customers
Regarding to the type and quality of service to be offered
Risk: penalty for failing to meet SLA
VoIP and voice over the Internet are not the same (What is the difference?)
SLA within the carrier network (Yes)
SLA between different carriers (is it possible?)

Front 9

Is it possible to have Qos end-to-end

Back 9

Only within a single carriers network.

In the US the internet is a collection of ISP's and is not possible to have end-to-end QoS service

Front 10

What are the approaches to QoS?

Back 10

Network over-provisioning - more bandwidth
Network Segregation
Separate Voice and Data Networks
IntServ and RSVP
Reserve the resource before establishing the session
DiffServ
Categorize traffic into different classes or priorities
Real-time applications with higher-priority values

Front 11

Network Segregation
Separate Voice and Data Networks

Back 11

logical separation of media and control

Front 12

What is RSVP?

Back 12

RSVP is the IP service that allows applications to request end-to-end QoS guarantees from the network. Cisco VoIP applications use RSVP for call admission control, limiting the accepted voice load on the IP network to guarantee the QoS levels of calls. The VoIP Call Admission Control using RSVP feature synchronizes RSVP signaling with H.323 Version 2 signaling to ensure that the bandwidth reservation is established in both directions before a call moves to the alerting phase (ringing). This ensures that the called party phone rings only after the resources for the call have been reserved. Using RSVP-based admission control, VoIP applications can reserve network bandwidth and react appropriately if bandwidth reservation fails.

Front 13

What is DiffServ?

Back 13

Behind all this success is the underlying fabric of the Internet: the Internet Protocol (IP). IP was designed to provide best-effort service for delivery of data packets and to run across virtually any network transmission media and system platform. The increasing popularity of IP has shifted the paradigm from "IP over everything," to "everything over IP." In order to manage the multitude of applications such as streaming video, Voice over IP (VoIP), e-commerce, Enterprise Resource Planning (ERP), and others, a network requires Quality of Service (QoS) in addition to best-effort service. Different applications have varying needs for delay, delay variation (jitter), bandwidth, packet loss, and availability. These parameters form the basis of QoS. The IP network should be designed to provide the requisite QoS to applications.
For example, VoIP requires very low jitter, a one-way delay in the order of 150 milliseconds and guaranteed bandwidth in the range of 8Kbps -> 64Kbps, dependent on the codec used. In another example, a file transfer application, based on ftp, does not suffer from jitter, while packet loss will be highly detrimental to the throughput.
To facilitate true end-to-end QoS on an IP-network, the Internet Engineering Task Force (IETF) has defined two models: Integrated Services (IntServ) and Differentiated Services (DiffServ). IntServ follows the signaled-QoS model, where the end-hosts signal their QoS needs to the network, while DiffServ works on the provisioned-QoS model, where network elements are set up to service multiple classes of traffic with varying QoS requirements. Both models can be driven off a policy base, using the CoPS (Common open Policy Server) protocol. Cisco IOS® Software supports both the IntServ and DiffServ models of QoS, along with an optional CoPS-client functionality

Front 14

Network Over Provisioning (No QoS)

Back 14

If the issue is congestion, we can over-provision the bandwidth by 200% or more.
With that, there would be no congestion issue.
Is it more cost-effective to do over-provisioning than to implement QoS on the network?

Front 15

Pro's and Con's of using bandwidth instead of QoS

Back 15

A simplistic but expensive solution
Simple: No major system development
Expensive: Significant overbuild
Unused for most of the time
An inefficient way
But, bandwidth is cheap and getting cheaper.
Is there really a need for QoS?
This debate will continue.

Front 16

Moore's Law

Back 16

Moore’s Law
Demand doubles roughly every 18 months
Bandwidth availability and bandwidth demand have tended to move almost in lock-step
New applications to use the available bandwidth
Lesson from the computer memory:
Is there a need for paging and other memory management techniques if we have unlimited computer memory?
Memory is never enough, so is the bandwidth.

Front 17

Advantages of Network Segragation

Back 17

The concern is that the data traffic may congest the network and cause the degradation of the voice service. If we separate voice and data networks, there will be no congestion issue.
Does it contradict the concept of converged networks?
It is about virtual segregation and not physical segregation.
Technologies
VLAN
Virtual Circuits (ATM and Frame Relay)
MPLS Label Paths

Front 18

IEEE 802.1p: LAN Layer 2 QoS/CoS Protocol for Traffic Prioritization

Back 18

IEEE 802.1p: LAN Layer 2 QoS/CoS Protocol for Traffic Prioritization
IEEE 802.1p specification enables Layer 2 switches to prioritize traffic and perform dynamic multicast filtering. The prioritization specification works at the media access control (MAC) framing layer (OSI model layer 2). The 802.1P standard also offers provisions to filter multicast traffic to ensure it does not proliferate over layer 2-switched networks.

The 802.1p header includes a three-bit field for prioritization, which allows packets to be grouped into various traffic classes. The IEEE has made broad recommendations concerning how network managers can implement these traffic classes, but it stops short of mandating the use of its recommended traffic class definitions. It can also be defined as best-effort QoS (Quality of Service) or CoS (Class of Service) at Layer 2 and is implemented in network adapters and switches without involving any reservation setup. 802.1p traffic is simply classified and sent to the destination; no bandwidth reservations are established.

The IEEE 802.1p is an extension of the IEEE 802.1Q (VLANs tagging) standard and they work in tandem. The 802.1Q standard specifies a tag that appends to an Ethernet MAC frame. The VLAN tag has two parts: The VLAN ID (12-bit) and Prioritization (3-bit). The prioritization field was not defined and used in the 802.1Q VLAN standard. The 802.1P defines this prioritization field.

IEEE 802.1p establishes eight levels of priority. Although network managers must determine actual mappings, IEEE has made broad recommendations. The highest priority is seven, which might go to network-critical traffic such as Routing Information Protocol (RIP) and Open Shortest Path First (OSPF) table updates. Values five and six might be for delay-sensitive applications such as interactive video and voice. Data classes four through one range from controlled-load applications such as streaming multimedia and business-critical traffic - carrying SAP data, for instance - down to "loss eligible" traffic. The zero value is used as a best-effort default, invoked automatically when no other value has been set.

Front 19

What is the QoS mechanism on the LAN.

Back 19

Networ Segragation using VLAN (802.1P and 802.1Q) to configure virtual LAN's on different subnets

Front 20

What is the QoS mechanism on the WAN?

Back 20

Multiprotocol Label Switch (MPLS) can dynamically configure separate label paths for voice and data traffic.
Different IP subnets for voice and data traffic
The bandwidth for the voice traffic can be pre-provisioned and the data traffic will not interfere (or congest) the voice traffic.
This is the same as Constant Bit Rate (CBR) for ATM virtual circuits.

Front 21

Is there a need for QoS?

Back 21

Yes. because there is network congestion

Over Provisioning
Network Segragation
RVSP
DiffServ

Front 22

The design of any network involves seeking a balance between four requirements:

Back 22

The design of any network involves seeking a balance between four requirements:
Meeting the requirements of the traffic demands (capacity): no blocking
Meeting the requirements of service features (features)
Minimizing the capital and operational cost of the network (cost)
Ensuring high network reliability and availability (quality)

Front 23

Design Criteria

Back 23

Traffic Analysis and Models
Busy Season Busy Hour (BSBH)
The busiest clock hour of the busiest week of the year
Scalability
Avoiding constant redesigning due to increased traffic
Technology Selection
SIP or H.323
MGCP vs. MEGACO/H.248
SS7 (legacy, SIGTRAN, SIP-T, etc.)
Reliability/Redundancy: network level and service level

Voice codec Selection
Packetization interval (delay)
Silence suppression
Call Admission Control (CAC)
Blocking: a call request is rejected due to a lack of available bandwidth
Is this applicable to VoIP? Who decides to block a call? What are the criteria to block a call?
QoS Consideration
Dedicated Bandwidth – network segregation
QoS at layer2 (802.1p)
Others

Front 24

Show where BHC/BHCA and Erlang calcualtions are used in the network

Back 24

Front 25

What is Traffic Modeling

Back 25

Based on the traffic demand to calculate the resources to determine the traffic quality

Front 26

Demand, Resources, Quality

Back 26

DEmand is about the callers, Resource is about the trunks, quality is about blocking

Front 27

What is BHCA

Back 27

In telecommunications, busy hour call attempts (BHCA) is a teletraffic engineering measurement used to evaluate and plan capacity for telephone networks. BHCA is the number of telephone calls attempted at the busiest hour of the day (peak hour), and the higher the BHCA, the higher the stress on the network processors. BHCA is not to be confused with busy hour call completion (BHCC) which measures the throughput capacity of the network. If a bottleneck in the network exists with a capacity lower than the estimated BHCA, then congestion will occur resulting in many failed calls and customer dissatisfaction.

BHCA is usually used when planning telephone switching capacities and frequently goes side by side with the Erlang unit capacity calculation. As an example, a telephone exchange with a capacity of one million BHCA is estimated to handle 250,000 subscribers. The overall calculation is more complex however, and involves accounting for available circuits, desired blocking rates, and Erlang capacity allocated to each subscriber.

Front 28

What needs to be determined in the Network Topology?

Back 28

How many network elements of a given type will be in each location
Media Gateway (MG)
Media Gateway Controller (MGC)
Technically, a single MGC could serve all MGs on the network, regardless of the location
SIP Proxy Server
The bandwidth requirements between those network elements and the outside world
How can we allocate bandwidth between network elements (MGs) where the network is a shared resource?

Front 29

What are the parameters used in the Eralng B Model

Back 29

Erlang-B Model
Traffic Intensity
Blocking Probability
Number of trunks

Front 30

MGC Quantities and Placement

Back 30

A call passes between two MGs controlled
By the same MGC
By different MGCs
Factors to determine MGC:
Call Capacity of MGC
Busy Hour Calls (BHC)
Number of SS7 and SIP messages that a MGC could handle
Delay
Requirements of SS7 interoffice delay: <150ms
Duration: from an ISUP message entering into an ingress MGC to the ISUP message exiting an egress MGC

Front 31

Objectives of Traffic Engineering

Back 31

Users can get dial-tone.
Line Origination
Users can make a phone call.
Signaling capacity
Switch Capacity
Trunk Capacity
Line Termination
What are the reasons that a call fails?
Bottleneck, congestion, etc.
Where?
review the previous diagram

Front 32

What are the limiting resources on the network?

Back 32

Front 33

There are three types of traffic flows. When there is no resource there are three possibilities;

1) traffic will be blocked and removed from the network

2) traffic will be blocked and then immediatel will retry

3) the traffic will put in the queue until a resource becomes available

What are the Erlang models called for each?

Back 33

1) traffic will be blocked and removed from the network.

This is called the Erlang B model

2) traffic will be blocked and then immediatel will retry

This is called the extended Erlang B model


3) the traffic will put in the queue until a resource becomes available. this is called the Erlang C model

Hint 33

In this class we focus on Erlang B where "Blocked Calls are Cleared from the System"

Front 34

There are three types of traffic flows. When there is no resource there are three possibilities;

1) traffic will be blocked and removed from the network

2) traffic will be blocked and then immediatel will retry

3) the traffic will put in the queue until a resource becomes available

What are the Erlang models called for each?

Back 34

1) traffic will be blocked and removed from the network.

This is called the Erlang B model

2) traffic will be blocked and then immediatel will retry

This is called the extended Erlang B model


3) the traffic will put in the queue until a resource becomes available. this is called the Erlang C model

Hint 34

In this class we focus on Erlang B where "Blocked Calls are Cleared from the System"

Front 35

What defines the Grade of Service?

Back 35

Blocking Probability

Blocking Systems
Call request is blocked (i.e. by fast busy signal) if no trunks are available.
GoS = Blocking Probability

Front 36

What are three Blocking System Examples

What are thre three parameters?

Back 36

Resource
Traffic
GoS

Front 37

What is Holding Time?

Back 37

Holding time = the length of time that a resource is being held (e.g. length of time for a phone call)
What is the average holding time of a residential call?
What is the average holding time of a business call?
What is the average holding time to any Tech Support?
What is the average holding time for an Internet call?

Front 38

What is Traffic Volume

Back 38

Traffic Volume for an interval = the sum of all the traffic holding times (of all calls) for that interval

Front 39

What is Traffic Intensity?

Back 39

Traffic Intensity = traffic volume / time interval which is a measure of demand
Traffic Intensity has direction (incoming traffic vs. outgoing traffic)

Front 40

Define Traffic Measurement

Back 40

Front 41

Calculate the erlangs for this example

Back 41

If the usage is in minutes divide by 60. If the usage is in seconds divide by 3600

Front 42

Describe BSBH, HDBH, and ABSBH

Back 42

Circuit traffic is collected for each hour of each day by the Switch’s SPC software. The data is averaged and reported in the following formats:
Busy Season Busy Hour (BSBH):
The busiest clock hour of the busiest week of the year
The engineering period for applying the GOS criteria.
High Day Busy Hour (HDBH)
Average traffic during the busiest day of the 10 days (excluding unusual days) in the highest traffic hour.
Average Busy Season Busy Hour (ABSBH)
The average of 20 BSBH
This measurement is applied to trunk groups between COs

Front 43

What is A,N,B,C, and U

Back 43

Front 44

For this example provide the following;

Offered Load
Blocking Probability
Carried Load
Circuit Utlization

Back 44

Front 45

For this example provide the following;

B, N, A, and C

Back 45

Front 46

For this example provide the following;

B, N, A, and C

Back 46

Front 47

What is the concept of CAC for VoIP?

Back 47

CAC is needed to ensure quality for accepted calls and block calls when quality cannot be ensured.

Front 48

How is CAC implemented?

Back 48

Performance metrics are required to monitor the network quality. When a performance metric (or a collection of them) is above a thresh-hold value, an incoming call request should be rejected (i.e., a call is blocked.)
This function can be implemented on Call Manager.
There are no standards for CAC for VoIP. Can you define a few performance metrics for CAC? What are the thresh-hold values to be used for these metrics?

Front 49

What is variable in CAC is the same concept as trunks?

Back 49

the max number of simultaneous calls

Front 50

Describe Call Admission Control (CAC)

Back 50

The network (call manager or softswitch) accepts a call request only if it could guarantee the quality of service (QoS) of the call.
In a network with dedicated bandwidth for VoIP, we can calculate the max number of simultaneous calls based on the allocated bandwidth.
This is the parameter N of the Erlang-B model
Maximum Call Load
When there are N calls in the network, any new call request will be rejected –
Same as no trunks are available to route the call.

Front 51

How do we caculate bandwidth?

Back 51

VoIP Bandwidth
Voice packet size + 40 octets (for IP, UDP and RTP) + layer 2 overhead
RTCP bandwidth should be limited to << 5% of the actual VoIP bandwidth.