Networks and Communication Lecture 1: Introduction

1
Networks and CommunicationLecture 1 Introduction

• Peter Steenkiste
• School of Computer Science
• Carnegie Mellon University
• ECOM, Summer 2000

2
Todays Lecture

• Course outline and goals.
• Basic networking introduction.
• Packet versus circuit switching.
• Application requirements.

3
Course Goals

• Become familiar with the principles and practice
  of networking and communications.
• Routing, transport protocols, naming, ...
• Get a basic understanding of some of the more
  commonly used technologies.
• Terminology, characteristics, future potential,
  ...
• Understand the potential and limitations of
  todays and tomorrows network technology.
• Performance, quality of service, security, ...

4
Course Format

• 14 lectures.
• Cover the material
• I will start from scratch (no networking
  background)
• Readings are posted beforehand
• Evaluation based on homeworks, midterm,
  programming assignment, and a final.
• Details on grade distribution later in course
• Course web page is used for information
  distribution.
• Readings, lecture notes, deadlines,
• http//www.cs.cmu.edu/prs/ecomm

5
Course Format

• Two teaching assistants.
• Servio Lima
• Suray Vasanth
• My office hours Wednesday 9-1030
• Wean Hall 8202
• TA office hours TBD.
• Traveling May 9-18, May 24-26.
• Guest lecturer for next two weeks

6
Policy on Collaboration

• Working together is important.
• Studying together, discuss course material in
  general terms, ...
• Work together on programming debugging, ..
• Homeworks and other individual assignments must
  be done individually.
• Feedback to you and to the instructor
• Collaboration defeats the purpose of the
  assignment
• Copying cheating
• See university policy on academic integrity
• Projects are sometimes done as a group.

7
Course Contents

• Basics.
• Requirements, protocol stacks, ...
• LAN and WAN technologies.
• Hardware, link layer protocols, wireless,
  throughput, ..
• Internetworking.
• Naming, addressing, routing, ..
• Network management.
• SNMP, ...
• Transport layers.
• UDP, TCP, video streaming, ..
• Quality of service.
• Principles, QoS models, signaling, RSVP, ..

8
A Single Network

• A set of switches, links and hosts that is
  managed as a single infrastructure.
• Also called an autonomous system or
  administrative domain
• Examples Andrew, MCI network,
• A (single) network is typically administratively
  fairly homogeneous.
• Same policies apply, same set of protocols are
  used, ..
• Managed by the same organization
• Typically no charges inside for internal
  communication
• Can physically very diverse, e.g. wired and
  wireless components

9
An Internetwork
Network
Network
Network
Network
Network
Network
Network
Network
10
The Internet

• An inter-net a network of networks.
• A set of networks that are connected with each
  other
• Networks are connected using routers that support
  communication in a hierarchical fashion
• Often need other special devices at the
  boundaries for security, accounting, ..
• The Internet the interconnected set of networks
  of the ISPs proving data communications services.
• In order to inter-operate, all participating
  networks have to follow a common set of rules.

11
Types of Networks

• Internet Service Providers (ISPs).
• Provide network service to customers
• Typically connect networks together
• ISPs can be international, national, regional, or
  local
• Access networks connect end-points.
• Can provide local connectivity or simply provide
  a path to an ISP
• Local-area network, dial-up through modem, ..
• Networks form a hierarchy.
• The further you have to go, the deeper you travel
  in the hierarchy.

12
Local Area Networks

• Specialized design opportunities
• Exploit small size (low latencies).
• Exploit traffic locality (don't saturate
  backbone).
• Exploit addressing/ routing locality (hierarchy
  of name servers).
• Exploit network management locality
  (administration, chargeback simpler).

13
Wide Area Networks

• Characteristics
• Longer physical delays.
• Higher degree of traffic aggregation.
• Supports many more hosts.
• More administrative diversity.
• Consequences
• Overall a more decoupled organization.
• More expensive equipment and transmission
  facilities - tighter monitoring of resource use.
• Scalability one of the main performance
  properties.

14
Protocols
Can you give me some directions?

• An agreement between parties on who communication
  should take place.
• Protocols may have to define many aspects of the
  communication.
• Data encoding, language, error recovery,
  termination conditions, ..
• Network protocols can exist between computer
  programs or hardware components.

Certainly, where would you like to go?
Heinz Hall
Go left at the next light
Thank you
15
More on Protocols

• Protocols are the key to interoperability.
• The hardware/software of communicating parties
  are often not built by the same vendor
• Sun workstation with PC, 3COM with Cisco bridge,
  ..
• Yet they can communicate because they use the
  same protocol
• Protocols exist at many levels.
• Application level protocols, e.g. access to mail,
  distribution of bboards, web access, ..
• Protocols at the hardware level allow two boxes
  to communicate over a link, e.g. the Ethernet
  protocol
• Intermediate protocols can add value to a
  lower-level protocol, e.g. provide a reliable
  communication service over an unreliable network

16
Packet Switching

• Source sends information as self-contained
  packets that have an address.
• Source may have to break up single message in
  multiple
• Each packet travels independently to the
  destination host.
• Routers and switches use the address in the
  packet to determine how to forward the packets
• Destination recreates the message.
• Analogy a letter in surface mail

17
Circuit Switching

• Source first establishes a connection (circuit)
  to the destination.
• Each router or switch along the way may reserve
  some bandwidth for the data flow
• Source sends the data over the circuit.
• No need to include the destination address with
  the data since the routers know the path
• The connection is torn down.
• Example telephone network.

18
Circuit Switching versusPacket Switching

• With circuit switching, the sender has to wait
  until the connection has been established before
  sending.
• Once the connection exists, communication can be
  very efficient
• Unused bandwidth on links is wasted
• Can have added value features, e.g. reservations
• Communication can start immediately with packet
  switching.
• But routers have to find the path for every
  packet
• More efficient for bursty traffic, but no
  reservations
• But there is more to the story
• Let us look at both technologies in more detail

19
Multiplexing

• With traditional circuit switching, the circuit
  completely occupies each wire along the path.
• There is an end-to-end physical circuit
• This can be very inefficient.
• Need many wires!
• Each user does not necessarily need the full
  bandwidth of the link, so there may be a lot of
  idle time
• The solution is multiplexing multiple users
  share the capacity of a link.

20
Frequency- versus Time-division Multiplexing

• With frequency-division multiplexing different
  users use different parts of the frequency
  spectrum.
• I.e. each user can send all the time at reduced
  rate
• Example roommates
• With time-division multiplexing different users
  send at different times.
• I.e. each user can sent at full speed some of the
  time
• Example a time-share condo
• The two solutions can be combined.

Frequency
Frequency Bands
Slot
Frame
Time
21
Virtual Circuits

• Circuit consists of a slice of the bandwidth on
  each link.
• Other slices are used by other users
• Different circuits can have different bandwidth.
• Must be multiple of some base bandwidth
• Switches have to remember that they should
  forward data from the red slot on link 5 onto the
  yellow slot of link 7.
• Still no need to carry addresses with the data

22
A Closer Look at Packet Switching

• Packet switches use store and forward
  transmission.
• Store packet, look up destination, forward packet
• Introduces a store-and-forward delay
• Sharing on a link is similar to time sharing,
  with two differences
• Destination is determined by address, not slot
  number
• Multiplexing is statistical, i.e. packets are
  interleaved without a fixed pattern

A
B
A
C
B
D
23
Circuit Switching versusPacket Switching
Revisited

• Circuit switching data flows over circuit.
• Delay of connection set up but data forwarding is
  fast
• Very predictable performance because of
  reservation
• Predictable bandwidth, no queuing delay, ..
• Unused bandwidth is wasted
• Very bad for bursty traffic, e.g. web browsing
• Packet switching forwarding based on address.
• No connection set up delay but store-and-forward
  delay
• Performance is unpredictable because of dynamic
  bandwidth sharing and lack of reservations
• Packet loss, unpredictable queuing delay, ..
• Can achieve high link utilization
• Other applications can use unused bandwidth
• Best solution depends on traffic load.

24
Pure PS versus Pure CS

• Every data packet carries a full destination
  address.
• No start up delay but per-packet processing
  overhead can be high.
• Dynamic bandwidth sharing results in
  unpredictable performance.
• Dynamic bandwidth sharing uses bandwidth
  efficiently.
• Little state in network.

• Data flows over cells based on at most a short
  circuit identifier.
• Delay associated with circuit setup but data
  forwarding is fast.
• Reserved bandwidth results in predictable
  performance.
• Static bandwidth allocation results in wasted
  bandwidth.
• A lot of state in network.

25
Intermediate Solutions

• Virtual circuits with statistical multiplexing.
• Bandwidth is divided in fixed-sized slots
• Each cell carries a virtual circuit identifier
• Switch uses virtual circuit identifier to forward
  cell
• Example ATM supports circuits both with and
  without reservations
• Packet switching with reserved bandwidth.
• User sets up an end-to-end connection with a
  certain reservation
• Packets still carry the address of the
  destination host
• Routers identify the packets belonging to that
  user and make sure that the user gets its share
  of the bandwidth
• Example IP is mainly best effort but supports
  for reservations is being introduced (slowly)

26
Spectrum of Options
Telecommunications Networks
Note naming slightly different from book
27
Network Service Model
host
host
host
host
host
host
host

• Core network responsible for transferring data
  between a sending and receiving host.
• End-to-end protocols present a network service
  to applications and users.
• May add value to the core network protocols

28
Protocol andService Levels
Application
End-to-end
Core Network
29
What Do Applications Care About?

• Predictability.
• Best effort document transfer
• Reservations telephony, video
• Reliability.
• Fully reliable various types of documents
• Some errors ok video
• Throughput.
• High bandwidth large images, high resolution
  video
• Low bandwidth telnet
• Delay.
• Delay sensitive telephony, interactive video
• Not delay sensitive document transfer

30
Network Service Models

• Set of services that the network provides.
• Best effort service network will do an honest
  effort to deliver the packets to the destination.
• Usually works
• Guaranteed services.
• Network offers (mathematical) performance
  guarantees
• Can apply to bandwidth, latency, packet loss, ..
• Preferential services.
• Network will give preferential treatment to
  packet in some packet classes, relative to other
  packet classes
• E.g. lower queuing delay
• Today no performance guarantees.
• Some differentiated service is being introduced

31
Reliability of the Data Transfer

• Core network offers a best effort service.
• Packets usually make it
• End-to-end protocols can add value
• Reliable connection-oriented service using the
  Transmission Control Protocol (TCP).
• User establishes (end-to-end) connection
• Service consists of a reliable bit pipe
• TCP deals with any errors in the core network
  service
• Unreliable datagram service using the User
  Datagram Protocol (UDP).
• Connectionless service
• User has to deal with any errors in the core
  network

32
Packet Delay

• Sum of a number of different delay components.
• Propagation delay on each link.
• Proportional to the length of the link
• Transmission delay on each link.
• Proportional to the packet size and 1/link speed
• Processing delay on each router.
• Depends on the speed of the router
• Queuing delay on each router.
• Depends on the traffic load and queue size

A
B
A
C
B
D
33
Sustained Throughput

• When streaming packets, the network works like a
  pipeline.
• All links forward different packets in parallel
• Throughput is determined by the slowest stage.
• Called the bottleneck link
• Does not really matter why the link is slow.
• Low link bandwidth
• Many users sharing the link bandwidth
• High processing times

50
267
17
37
59
30
104
34
Application-level Delay
Delay of one packet
Average sustained throughput
For minimum sized packet
35
Other Requirements

• Network reliability.
• Network service must always be available
• Security mechanisms privacy, access control, ..
• Fairness.
• Rich functionality.
• Scalability.
• Scale to large numbers of users, traffic flows,
  ...
• Manageability monitoring, control, ..
• Requirement often applies not only to the core
  network but also to the servers.
• Requirements imposed by users and network
  managers.

36
Bandwidth Sharing

• Bandwidth received on the bottleneck link
  determines end-to-end throughput.
• Router before the bottleneck link decides how
  much bandwidth each user gets.
• Users that try to send at a higher rate will see
  packet loss
• User bandwidth can fluctuate quickly as flows are
  added or end, or as flows change their transmit
  rate.

BW
100
Time
37
Fair Sharing of Bandwidth

• All else being equal, fair means that users get
  equal treatment.
• Sounds fair
• When things are not equal, we need a policy that
  determines who gets how much bandwidth.
• Users who pay more get more bandwidth
• Users with a higher rank get more bandwidth
• Certain classes of applications get priority

BW
100
Time
38
Wireless and Mobile Networks

• Wireless communication based on electro magnetic
  waves traveling through the ether.
• No need for a wire to carry the signal
• More error prone and lower data rates
• Mobile communication computer moves around in
  the network.
• Complicates addressing
• Wireless and mobile communication are different
  issues.
• Wireless mobile moving around with a wireless
  laptop
• Wireless only wireless connections to isolated
  areas
• Mobile only traveling with an Ethernet capable
  laptop
• Disconnected or intermittently connected
  operation.