View
4
Download
0
Category
Preview:
Citation preview
1Switching, impairments, metrics ...
Network Switching Technologies, Impairments, and Metrics
#3
2Switching, impairments, metrics ...
Outline
� Basic networking technologiesCircuit SwitchingPacket Switching Statistical Multiplexing
Virtual Circuit Packet Switching� Network impairmentsHow the physical environment impacts network protocols
� Performance Metrics
3Switching, impairments, metrics ...
Network Switching Technologies
Input port 2Line Card B
Line CardInterface
Input Lines
Output Lines
Output port 3Line Card C
Switch Fabric
Control
A
B
C
D
A
B
C
D
2
ArrivingPackets
Buffer(Queue)
Server (b/s)
DepartingPackets
Output port for packet switches
4Switching, impairments, metrics ...
Controller
1
2
3
N
Line card
Line card
Line card
Line card
Inte
rcon
nect
ion
fabr
ic
Line card
Line card
Line card
Line card
1
2
3
N
Input ports Output ports
Data path Control path
…………
Generic Packet Switch
“Unfolded” View of Switch� Ingress Line Cards
Header processing Demultiplexing Routing in large switches
� Controller Routing in small switches Signaling & resource
allocation� Interconnection Fabric
Transfer packets between line cards
� Egress Line Cards Scheduling & priority Statistical Multiplexing
Modified from: Leon-Garcia & Widjaja: Communication Networks
5Switching, impairments, metrics ...
Inte
rcon
nect
ion
fabr
ic
Transceiver
Transceiver
Framer
Framer
Networkprocessor
Backplanetransceivers
Tophysicalports
Toswitchfabric
Tootherlinecards
Line Cards
Folded View� 1 circuit board is ingress/egress line card
- Physical layer processing- Data link layer processing- Network header processing- Physical layer across fabric + framing
Modified from: Leon-Garcia & Widjaja: Communication Networks
Line card
6Switching, impairments, metrics ...
� Switching transfers information from input ports to output ports
� Elements of Switches Line cards (multiple interfaces/card) Switch fabric Control (mappings) Management Billing
Network Switching Technologies
Input port 2Line Card B
Output port 3Line Card C
Dual power supplies.Why?
7Switching, impairments, metrics ...
Network Switching Technologies: Circuit Switching
� On demand call set up� Dedicated fixed path� Fixed network resources held for call duration
End-Office
Long-distanceOffice
Long-distanceOffice
End-OfficeTrunksLocal loop
8Switching, impairments, metrics ...
Network Switching Technologies: Circuit Switching
� Phases of a callCall establishment Information transferCall disconnect
� If no network resources are available the call blocks (fast busy signal)� Network resources can be easily defined:Voice lineWavelength (color) in DWDM systems
9Switching, impairments, metrics ...
Network Switching Technologies: Circuit Switching
� Examples of circuit switching Legacy Public Switched Telephone Network (PSTN)Dense Wavelength Division (DWDM) Optical Networks
� Customers can not use idle channels, unused capacity is wasted
10Switching, impairments, metrics ...
Network Switching Technologies: Message Switching
� No Dedicated path� Address information is added to the message� Store if output port is busy� Trade off delay for blocking� If message is corrupted then retransmit entire
message
Store&
Forward
Store&
Forward
Store&
Forward
ArrivingMessages
Buffer(Queue)
Server (b/s)
DepartingMessages
11Switching, impairments, metrics ...
Network Switching Technologies: Message Switching
� If message arrives to empty system then it is transmitted at the FULL LINE RATE
� Transmission at the FULL LINE RATE is statistically shared among the all the users
� If a message arrives to a busy system it waits
Buffer Server
Message
This is called a Statistical Multiplexer
12Switching, impairments, metrics ...
Network Switching Technologies: Packet Switching
� Break up large messages into smaller units: Packets� The process of “breaking up” larger information units into smaller
parts is called: Segmentation or Fragmentation� The process of “putting together” smaller parts into larger
information units is called: Reassembly� Segmentation and Reassembly (SAR) can happen multiple times
to the same information stream, or flow
13Switching, impairments, metrics ...
TDM vs Statistical Multiplexing
Server
Dwell time = one time slotUser 1
User N
Server
User 1
User N
TDM
Statistical Multiplexing
If system empty then immediately TX packet at total link rateDelay wait for turn and then TX packet at total link rateLoss arrive and buffer is full
Link Rate in b/s
Link Rate in b/s
If all time slots are allocated upon request for transmission then the new “call” is blocked
14Switching, impairments, metrics ...
TDM vs Statistical Multiplexing� Example:
Link rate = 1 Mb/s Message size = 5000 bits TDM
– Frame size = 10 ms ( 1 Mb/s * 10 ms =10,000 bits/frame)
– 10 times slots /frame ( slot time = 1 ms, 1000 bits/slot)
– Time to transmit 5000 bit message = 5 slot times @ 1 slot /frame = 50 ms
Statistical Multiplexing– “Assume” arriving message finds the system empty– Time to transmit 5000 bit message =
5000 bits/(1Mb/s) = 5 msWhen is TDM faster?
15Switching, impairments, metrics ...
Network Switching Technologies: Packet Switching
Packet Switch
Forwarding Table
Inpu
t Por
ts
Ou
tpu
t Por
ts
CPU
Input Port i@ Rate = Ri (b/s)
Output Port j@ Rate = Rj (b/s)
16Switching, impairments, metrics ...
Network Switching Technologies: Classes of Packet Switching
� Datagram Packet SwitchingA form of Conectionless service
� Virtual Circuit (VC) Packet SwitchingA form of Connection-oriented service
17Switching, impairments, metrics ...
Network Switching Technologies: Packet Switching --> Datagrams
� Each packet is treated independently� Packets with same destination may take different routes through the network� Each Network Element (NE), e.g., router, makes independent forwarding decisions� Packets can arrive out-of-order� No call set up is required� Nodes keep no “state” information at the network layer� No QoS is guaranteed
18Switching, impairments, metrics ...
Network Switching Technologies: Packet Switching -->Datagrams
To: John SmithAddress
From: Jane DoeAddress
User Data insideenvelope
19Switching, impairments, metrics ...
Network Switching Technologies: Packet Switching --> Datagrams
Store&
Forward
Store&
Forward
Store&
Forward
Destination Address Source AddressData
Packet
Packet Header (overhead)
Why is the source address included in the packet header?
20Switching, impairments, metrics ...
How do Packets find there way: Routing and Forwarding
1
23
0111
Value in arrivingpacket’s header
routing algorithm
local forwarding table
header value output link
0100010101111001
3221
Modified from Computer Networking: A Top Down Approach Featuring the Internet, 4nd edition. Jim Kurose, Keith Ross, Addison-Wesley, Copyright 1996-2002, J .F Kurose and K.W. Ross, All Rights Reserved
Value related to destination
The routing algorithm builds the forwarding table using
“other” knowledge
21Switching, impairments, metrics ...
Network Switching Technologies: Packet Switching --> Datagrams
�Datagram is an example of a connectionless service� Internet Protocol (IP) provides a connectionless
service
22Switching, impairments, metrics ...
Network Switching Technologies: Virtual Circuit Packet Switching
� Virtual circuit packet switch provides a connection oriented service� A “logical connection” is established between the source and destination using all the
NEs along the set-up path� All packets flow over the same route through the network� Packets still “statistically share” link� Connection oriented does not imply virtual circuits
23Switching, impairments, metrics ...
Virtual circuits: signaling protocols
� Used to setup, maintain teardown VC� Used in
DWDM networks Multiprotocol Label Switching MPLS Used in legacy technologies, ATM, frame-relay, X.25
applicationtransportnetworkdata linkphysical
applicationtransportnetworkdata linkphysical
1. Initiate call 2. incoming call3. Accept call4. Call connected
5. Data flow begins 6. Receive data
These are Packet Switches
Modified from Computer Networking: A Top Down Approach Featuring the Internet, 4nd edition. Jim Kurose, Keith Ross, Addison-Wesley, Copyright 1996-2002, J .F Kurose and K.W. Ross, All Rights Reserved
24Switching, impairments, metrics ...
Network Switching Technologies: Virtual Circuit Packet Switching
� Forwarding decisions are made based on a “virtual circuit identifier” (VCI)not on the full address
� Packet share transmission facilities� Switches save state/connection� State is saved for duration of the connection along the path� QoS can be guaranteed
25Switching, impairments, metrics ...
Network Switching Technologies: Virtual Circuit Packet Switching
*Note: Do not need the same VCI end-to-end
Switch 3
Port In VCI In Port Out VCI Out
7 5 1 55
NIU=Network Interface Unit
26Switching, impairments, metrics ...
Network Switching Technologies: Virtual Circuit/Datagram Trade-offs
Example: Find the time to transmit a 1 Kbyte message coast-to-coast is the USA (3000Km) on a 600 Mb/s link. Propagation time = 3000km/c=10ms
a) Using Datagrams: 1*8000/(600Mb/s) + 10 ms ~ 10ms
Clocking time =(13 us)
b) Using Virtual CircuitsA B
Call Set-up
Data Transmission
30ms
What is the Datagram vs VC performance trade-off considering propagation time, and holding time=service time=L(bits)/R(b/s)?
27Switching, impairments, metrics ...
Network Switching Technologies: Virtual Circuit/Datagram Trade-offs
Example: Find the time to transmit a 37.5 Mbyte message coast-to-coast is the USA (3000Km) on a 600 Mb/s link
a) Using Datagrams: 510ms
b) Using Virtual Circuits
530ms
Key issue is holding time relative to call set-up time
A B
Call Set-up
Data Transmission
28Switching, impairments, metrics ...
Comparison
Modified from: Computer Networks, A. S. Tannenbaum, 4th Ed, Prentice Hall, 2003
29Switching, impairments, metrics ...
Example: Message=1920B, Header=40B, Packet size=1000B
RouterB
RouterC
Host A Host D
RAB =800 Mb/s
AB = ~0 sec
RBC =80 Mb/s
BC = 10 ms= 10,000 s
RCD =800 Mb/s
CD = ~0 sec
DatagramA DCB
tfAB = tfCD = 8000bits/(800Mb/s)=10s
tfBC = 8000bits/(80Mb/s)=100s
Tim
e
10s20s
ss
s
s
s =10,220s
Virtual CircuitCall set up +10,220s = s+ s +10,220s = 30,220s
0s Clocking Times
30Switching, impairments, metrics ...
Network Switching Technologies:
� Permanent Virtual Circuits (PVC)VC numbers assigned by management, usually manual Permanent Virtual Circuits are always connected
� Switched Virtual Circuits (SVC)Automatic user initiated call set-upsCall set up is on-demand
31Switching, impairments, metrics ...
Network Switching Technologies: Switch (NE) State
�Circuit switched networksSwitches have “hard” state, they save knowledge of the
connection throughout the duration of the call�Pure datagram networks Routers (switches) have no state of each connection or
“flow”.Can use a ‘label’ inside the packet for priority queuing
32Switching, impairments, metrics ...
Network Switching Technologies: Soft state
� Often multiple related flows pass through a network element� Implicit identification of what datagrams constitute a flow can
be done based on source/destination addresses� The states can be used for resource management --
RSVP (Resource Reservation Protocol)� State of flows are temporarily saved� Unused soft state is cleared by time outs
33Switching, impairments, metrics ...
Network Switching Technologies: Attributes of datagram transmission
� Decentralized control� Simple hosts and switches� No call processing� Difficult to provide QoS� Difficult to bill for services (often default to GB/month)� Difficult to manage resources to control congestion, note if no congestion
then not an issue
34Switching, impairments, metrics ...
Network Switching Technologies:Attributes of Virtual Circuit Packet Networks
� Provides QoS� Accommodates billing� Centralized control� Follows well known model of the PSTN� Simpler switching� Complex control� Soft state is trying to give
connectionless systems attributes of an connection oriented network
35Switching, impairments, metrics ...
Datagram and VC networks
� Networks that only provide VC services at the network layer are typically called VC networks
� Networks that only provide connectionless services at the network layer are typically called datagram networks
36Switching, impairments, metrics ...
Network Impairments
� Propagation Delay: The Speed of Light Limitation
Propagation delay = Distance (m)/(Speed of Light m/s)
� Example: 3000 km fiber linkSpeed of light in fiber = 0.66*(3x108m/s)Propagation delay = 3000x103m/ 0.66*(3x108m/s)= 15ms
� +Other Media– Speed of light in free space= 1.0*(3x108m/s) – Speed of light in coax= 0.88*(3x108m/s)
� Effect of clocking time, L/R “putting the bits on the link”
PacketSource Destination
Clocking time = L/C = 1 ms
Example: Distance = 3000km, Data rate = 1 Mb/s, Packet size = 1000 bits
Propagation time = 15 ms
37Switching, impairments, metrics ...
Network Impairments: Propagation Delay
� PAN: Personal Area Networks [BAN: Body Area Network]
~ 3 m or 10ns� DAN.: Desk Area Networks
~ 3 m or 10ns� LAN: Local Area Networks
~3 Km or 10us
� MAN: Metropolitan Area Networks ~300 Km or 1ms
� NAN: National Area Networks ~3000 Km or 10ms
� GAN: Global Area Networks ~10,000 Km or 30ms NANs and GANs are typically called
WANs Wide Area Networks� Terrestrial Networks� Satellite Networks
For GEO~500ms
� Interplanetary Networks
38Switching, impairments, metrics ...
Network Impairments: Delay-Bandwidth Product
� Delay-Bandwidth Product One way propagation delay = sec Link rate = R b/s Delay-Bandwidth Product = 2 R bits
� Number of packets in Delay-Bandwidth Product L Packet length in bits/packet Number of packets in Delay-Bandwidth Product = Number of packets in round trip time (RTT) =Delay-Bandwidth Product/L
� Example: D= 2000 km c= 2x108m/s (fiber)
= 10 ms L = 1000 Bytes R = 10 Gb/s Delay-Bandwidth Product = 200 Mb # packets in DBP=25000 The transmission line is “storing” 25000
packets
39Switching, impairments, metrics ...
Network Impairments
� Error environmentWired/Fiber/CableWireless
40Switching, impairments, metrics ...
Network Impairments:Error Environment
� Model 1: Random, bit errors are independent
� Model 2: Bursty, bit errors are correlated and come in groups,
BER= 10-12 BER= 10-1
41Switching, impairments, metrics ...
Network Impairments:
� Example: Impact of delay and errors: Link rate 600 Mb/s Free Space Link distance 3000 km 10ms Packet size:
– Payload 48 bytes– Overhead 5 bytes– Total 53 bytes ( 424 bits )
42Switching, impairments, metrics ...
Network Impairments:
14,285 1
424 bits/(600Mb/s) = .7us/packet10ms/(0.7us/packet) = 14,285 packets in flight
Many packets can be in transit.
Question: How do you cope with packets in error?
43Switching, impairments, metrics ...
Network Impairments:
�Example:Line rate = 600 Mb/sBit error rate( BER ) = 10-9
� What is the time between errors:On average see one error in 109 bits (109 bits/error)/(600 Mb/s) = 1.66 sec between errors
44Switching, impairments, metrics ...
Network Impairments and Application Types
� Real time interactive applications Require fixed or bounded delays Large delay “variance” can degrade performance Some real-time applications can tolerate some errors, e.g.,
– Voice– Video
� Non Real time (elastic) applications Can tolerate delay variance Require accurate delivery of information Does not require ‘timely’ delivery of data Can not tolerate errors
– E-mail– Telnet– FTP– www
� Some applications do not fit nicely into these categories
45Switching, impairments, metrics ...
Network Performance Criteria: Response Time
Response time TR: The time to “correctly” transmit a packetfrom Source to destination.
“correctly” implies Response time includes acknowledgments
Source Host
Network Interface
Card (NIC)
Network
DestinationHost
Network Interface
Card
Examine key components of delay
46Switching, impairments, metrics ...
Network Performance Criteria:Response Time
� Time from source applications to NIC� Waiting time in NIC to enter the network: buffering time� Time to transmit the packet: clock the packet into the network� Time for the network to deliver the packet to the destination’s NIC� Time for destination’s NIC to generate an acknowledgment� Time for the acknowledgment to reach the source host: repeating the above
steps
47Switching, impairments, metrics ...
Network Performance Criteria:Response Time Dependencies
� State of the network Current topology Active nodes Active links
� State of the other users Congestion
� Errors� State of source/destination� Link speeds� Message sizes� Message priorities
� Response time, TR is a random variable Probability density
function characterizes TR
% packets observed with delays greater than T
Variance Mean
48Switching, impairments, metrics ...
Network Performance Criteria: Response Time
� Network designers focus on the components of response time that are a function of the network
� Find the delay as a function of: traffic load packet length topology
� We will find the average queueing delay, time from packet entering a buffer until the last bit is clocked out
49Switching, impairments, metrics ...
Network Performance Criteria: Throughput
� Throughput in b/s or packets/sec,� Normalized throughput
= Average error free rate (b/s) passing a reference point in the network
S = % time the network is carrying error free packets-goodput
C = Link Capacity (b/s) = Peak link rate (R)
RAve
50Switching, impairments, metrics ...
Network Performance Criteria
� Channel (or link) utilization: The % time the channel (or link is busy)
� Channel Efficiency The % time the channel is carrying user information
(impact of overhead)Example: Let
D = #user data bits / packet
H = #network overhead bits/ packet=number of bits in the header
thenChannel efficiency = S( D
D + H)
51Switching, impairments, metrics ...
Network Performance Criteria: Throughput
� Channel Capacity, Smax, is the maximum obtainable throughput over the entire range of input traffic intensities, i.e., offered load.
Normalized Offered Load
ThroughputIdeal Case
1
1
Maximum Throughput =smax
Maximum Lossless Throughput
52Switching, impairments, metrics ...
Network Performance Criteria:Case Study
From: ATM WAN performance tools, experiments, and resultsDaSilva, L.A.; Evans, J.B.; Niehaus, D.; Frost, V.S.; Jonkman, R.; Beng Ong Lee; Lazarou, G.Y.;Comm Magazine, IEEE Volume 35, Issue 8, Aug. 1997 Page(s):118 - 125
With 45 Mb/s (DS-3)Access Lines
53Switching, impairments, metrics ...
Network Performance Criteria: Reliability
� Reliability: The reliability of a network can be defined as the probability that the functioning nodes are connected to working links.
Reliability = 1 - Network Failure� Here lets assume all nodes are working and
analyze the “Reliability” of basic ring and tree networks where only links fail
54Switching, impairments, metrics ...
Network Performance Criteria: Reliability
4 links
5 links:every node has two paths
Tree NetworkTopology
Ring Network Topology
NE=Network Element
NE
NE
NENE
NE
NE
NE NE
NE
NE
55Switching, impairments, metrics ...
Other Network Topologies
Full Mesh NetworkTopology
Bus NetworkTopology
Note a wireless network can be viewed as a bus topology
All users hear all transmissions. Transmission media coordinated
using a MAC protocol.
56Switching, impairments, metrics ...
Other Network TopologiesBus Network Topology Key concepts:
1) Broadcast Everyone on the same “Network” can directly communicate, using a point-to-point link, DLC protocol.2) Packet delivered to one special node, i.e., a router, connected to the “Network” is sufficient for delivery to any destination connected to the broadcast network.
Router
Internet
57Switching, impairments, metrics ...
Network Performance Criteria: Reliability
� Reliability for a 5 node tree network� Any of the 4 links fail the network is down� Let p = probability of link failure and assume failures
are statistically independent � Then Prob[no link failure] = (1-p)4
� Prob[network failure] = 1 - (1-p)4
58Switching, impairments, metrics ...
Network Performance Criteria: Reliability
� But (1-p)4 = 1 - 4p + 6p2- 4p3 + p4
� Prob[network failure] = 4p - 6p2 + 4p3 - p4
� Assuming p is small then for 5 node tree network the Prob[network failure] 4p
59Switching, impairments, metrics ...
Network Performance Criteria: Reliability
� Reliability for a 5 node ring network� Ring network has 5 links� Ring network can have one link failure and still be working, note one more
link can fail� Let q = 1 – p = probability of link good� Prob[network good] = Prob[all good or one failed and 4 good] = Prob[all good] + Prob[ one failed and 4 good] =q5+ 5p q4
� So Prob[network failure] = 1 - q5 - 5p q4
Prob[all good]=q5 Prob[ one failed and 4 good] =
60Switching, impairments, metrics ...
Network Performance Criteria: Reliability
� Expanding Prob[network failure] = 10p2q3 + 10p3q2 + 5p4q +p5
� The dominant term (assuming p small) is 10p2q3
Tree Ringp 4p 10p2q3
0.01 0.04 0.000970.001 0.004 10-5
10-5 4x10-5 10-9
10-7 4x10-7 10-13
61Switching, impairments, metrics ...
Network Performance Criteria: Reliability
62Switching, impairments, metrics ...
Other Network Performance Criteria
� Blocking ProbabilityPacket, we will find the probability of packet lossCall/Session, we will find the probability that all
resources are busy when a request arrives� Fairness� Security
63Switching, impairments, metrics ...
Network Performance Criteria:Example
From: CAPOZZI et al.: DOWNLINK PACKET SCHEDULING IN LTE CELLULAR NETWORKS: KEY DESIGN ISSUES AND A SURVEY, IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO. 2
RRM = Radio Resource Management
Bits/Sec/Hz
64Switching, impairments, metrics ...
Network Performance Criteria:Example
QCI= Quality-of-Service Class Identifier
GBR = Guaranteed Bit Rate IMS = IP Multimedia Core Network Subsystem
From: CAPOZZI et al.: DOWNLINK PACKET SCHEDULING IN LTE CELLULAR NETWORKS: KEY DESIGN ISSUES AND A SURVEY, IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO. 2
65Switching, impairments, metrics ...
Network Performance Criteria:Example
From: DoD Proposal RFP October 1997
TrafficData Type
NetworkService
TypicalMessageSize
QoS SoS
BattleCommandData
IP Unicast&Mulitcast
100-500Bytes
0.5sec>95%
SituationAwareness
IPBroadcast
5 sec<100 Bytes >95%
QoS = Quality of Service = % IP packets successfully delivered (for this RFP)
SoS = Speed of Service = Average time to deliver error free packets
66Switching, impairments, metrics ...
Network Performance Perspective: User-Oriented
� Minimum application response time (Delay guarantee)
� Maximum application throughput (Throughput guarantee)� Low loss (Maximum packet loss guarantee)� Highly reliable (Availability guarantee)� Very flexible� Secure� Low cost
67Switching, impairments, metrics ...
Network Performance Perspective: Network Manager/Provider
� Maximum throughput for all users� Effective congestion control� Power = Throughput/Delay� Easy of management� Highly reliable� Fairness� Ease of billing� Low cost
68Switching, impairments, metrics ...
Network Performance Perspective: Network Designer/Developer/Vendor
� Simple design� Robust� Scales
Number of users Geographical distribution Speed
� Efficient use of resources, CPU, links and memory� Evolvable
69Switching, impairments, metrics ...
Network Performance: What Can the Network Guarantee?
�Quality of Service (QoS)- Absolute/Contractual performance guarantees Examples:
–Sustainable rate–Peak rate–Packet delay (average and standard deviation)–Packet/Cell loss rate
�Network must reserve resources to provide QoS
70Switching, impairments, metrics ...
� Class of Service (CoS)-Relative performance guarantees Examples: Best Effort (lowest priority) [Current core Internet is Best Effort]
– e-mail– ftp
Gold (medium priority)– Point of sales transaction
Platinum (highest priority)– Voice– Video
� Network performs packet ‘labeling’ and priority queueing to provide CoS� Differential Services (IP-DiffServ) provides CoS in the Internet
Or: Bronze Sliver Gold
Or: Bronze Sliver Gold Platinum
Network Performance: What Can the Network Guarantee?
71Switching, impairments, metrics ...
� Network engineering issues: QoS
– Reserving resources implies per connection processing and saving flow state information for each flow
– Reserving resources implies call set-up– Reserving resource thus implies complex per flow processing in each switch
CoS– Relative services implies simplified router processing only look at a label and
queue appropriately– Relative services implies no unique per flow processing
Hybrids– QoS in the enterprise and CoS in the core network– QoS in the core network and CoS in the enterprise
Network Performance: What Can the Network Guarantee?
72Switching, impairments, metrics ...
How to Bill for it?� Usage Tiers: Customers can choose from ISP provided
Date Allotment: GB per month Peak Upstream/Downstream Rates Not CoS per application
� Bandwidth Caps: All-you-can-eat — to a point. Requires a tool to let customers manage usage and give warnings when levels are close to being reached. Also includes the opportunity to buy additional capacity when needed.
� Bandwidth Credits: Like a cellular minutes rollover, how about crediting customers when they come in below their usage agreements, minimizing usage?
� Power Boost: Customers gain access to more bandwidth on an as-needed-basis —useful for one-time downloads or occasional requirements.
� Application-Specific Bandwidth: Price bandwidth reservations into specific applications, such as video-on-demand, carrier-provided or third-party VoIP, or online gaming.
Modified from:10 Ideas for Broadband Billing, Telephony, Oct 2008
73Switching, impairments, metrics ...
How to Bill for it?
� Third-Party Subsidies: Charges for broadband are subsidized by other parties: advertisers, third-party VoIP operators, over-the-top Web content providers.
� Ad-Supported: Users gain access to bandwidth — or more bandwidth — in exchange for viewing advertising or sharing profile information.
� User Controls: Consumption is managed by giving a responsible party — parent, corporate IT department — control via preset usage and spending limits. Alerting and notifications leads to self-management of consumption.
� Off-Peak Usage: Set prices and policies to encourage off-peak usage, such as overnight delivery of cached video.
� Priced-Right All-You-Can-Eat: There's almost always a price at which unlimited access makes economic sense for carriers, aided by the fact that some unlimited customers will under-consume significantly, subsidizing biggest consumers.
Modified from:10 Ideas for Broadband Billing, Telephony, Oct 2008
Recommended