Yelena Yesha

1

Networking Technologies

• Yelena Yesha
• Olga Streltchenko

2
Presentation Overview

• Evolution of Networks.
• Networking Challenges.
• Types of Networks.
• Network Principles.
• Internet Protocols.
• Summary.

3
The Network

• Built from
• Transmission media
• Wire, cable, fibre, wireless channels
• Hardware devices
• Routers, switches, bridges, hubs, repeaters,
  network interfaces
• Software components
• Protocol stacks, communication handlers, drivers.

4
Evolution of Networking

• Batch Environment - 1950s
• no direct interaction between users and their
  programs during execution.
• Time Sharing - 1960s
• Dumb terminals were connected to a central
  computer system.
• Users were able to interact with the computer and
  could share its information processing resources.
• Marked the beginning of computer communications.

5
Evolution of Networking (cont'd)

• Distributed Processing use of minicomputers -
  1970s
• Users demanded computing closer to their work
  areas.
• Communication between neighbour processors and
  applications via networks.
• WAN and LAN- 1980s
• Internet, broadband and wireless communication,
  mobile code, ubiquitous computing, etc. - 1990s
• 2000s - ?

6
Networking Challenges

• Performance
• Scalability
• Reliability
• Mobility
• Security
• QoS (Quality of Service)

7
Performance

• Parameters that determine the speed of message
  exchange between two nodes
• Latency
• Delay that occurs after a send operation and
  before the data becomes available at the target
  node, i.e. latencytime to transmit an empty
  message
• Data transfer rate
• The speed at which data can be transferred
  between two nodes (bits/sec).
• If a message length does not exceed the max
  determined by the network technology, then
  Message transmission timelatencylength/data
  transfer rate

8
Performance (cont'd)

• Transfer rate is primarily determined by physical
  characteristics of the network.
• Latency is primarily determined by
• software overheads,
• routing delays,
• load-dependent non-deterministic elements
• E.g., message collision on the Ethernet.
• Total system bandwidth of a network
• Measure of throughput
• Total volume of traffic that can be transferred
  across the network in a given time.

9
Scalability

• A system is described as scalable if it remains
  effective when there is a significant increase in
  the number of resources and the number of users.
• Challenges in scalable system design
• Controlling the cost of physical resources as the
  demand for resources grows
• e.g., for a system with n users the quantity of
  physical resources should be at most O(n).
• Controlling the performance lost as the number of
  users/resources grows
• e.g., for a system with n objects the access time
  should be at most O(log n).

10
Scalability (cont'd)

• Challenges in scalable system design (cont'd)
• Preventing software resources running out
• Example 32-bit IP address of the 1970's ran out
  current IP address uses 128 bits and is expected
  to be exhausted by early 2000's. - Keeping up is
  a serious challenge!
• Avoiding performance bottlenecks
• Use decentralized algorithms, caching,
  redundancy and replication
• Example DNS table maintenance tables are
  distributed and replicated.

11
Scalability on the Internet

• Potential size of the Internetworld population.
• Original network technologies did not anticipate
  this scope.
• Changes to the addressing and routing.
• Current average round-trip time 100-150ms
• Individual numbers vary widely.
• The ability to scale will depend on the economics
  of use
• Charges to the users
• Patterns of communication.

12
Reliability Failure Models

• Communication failures (vs process failures)
• Omission failure communication channel fails to
  perform prescribed actions
• e.g., loss of messages
• Easiest type of failure to detect and handle,
  e.g., retransmit the message.
• Arbitrary failure unintended actions occur (any
  type of error)
• e.g., delivery of a corrupted message, delivery
  of a non-existent message, repeated delivery
• This type of error is rare since communications
  software is able to detect and correct it.

13
Reliability Failure Models (cont'd)

• Communication failures (cont'd)
• Timing failure arises in synchronous application
  where time limits are set on message delivery
• Responses become unavailable to clients after
  timeout, e.g., ftp
• Asynchronous systems like WWW are not suseptible
  to this type of error since they do not provide
  any timing guarantees.

14
Handling failures

• Detecting
• E.g., use checksum to detect a corrupted message
• Not always possible, e.g., a remote server crash.
• Masking
• Hide a failure
• By means of service/data replication, etc.
• Convert a failure into another type of failure
• e.g., dropping a corrupted message turns an
  arbitrary failure into an omission failure
• We know how to handle it.

15
Handling Failures (cont'd)

• Tolerating
• Impractical to detect and hide all the failures
  on the Internet
• Software informs users about failure
• Include redundant components into the system to
  tolerate failures, e.g.
• at least two different routes between two
  routers
• DNS replication
• operational database replication.

16
Handling failures (cont'd)

• Recovery
• Involves special software design that allows to
  recover the state of the permanent data.

17
Reliability of Communications Requirements

• Validity
• Any message in the outgoing buffer will be
  eventually delivered to the incoming message
  buffer.
• Integrity
• The message received is identical to the message
  sent, and no messages are delivered twice.

18
Mobile Code

• Code that can be sent from one computer to
  another
• e.g., Java applets
• Virtual Machine approach
• A way of making code executable on any hardware
• VM is middleware, i.e. a layer of software whose
  purpose is to mask heterogeneity of hardware
• The compiler generates code for a VM
• Used by Java and is not necessarily extendable to
  other languages.

19
Mobile Code (contd)

• The advantage of running downloaded code is
  network delay avoidance during interactions.
• Potential security threat to the local resources.

20
Mobile Agents

• A running program (code and data) that travels
  from one computer to another over the network
  carrying out a task on behalf of a user
• e.g., to perform information retrieval.
• The advantage over client-server approach lies in
  the reduction of communication time and cost
• replaces remote invocations with local ones.
• Potential security threat to the host.
• MA are vulnerable themselves.

21
Mobile Devices

• Proliferation of small and portable computer
  devices
• e.g., laptops, PDAs, mobile phones, digital
  cameras, etc.
• Enabled with wireless networking
• Metropolitan or greater ranges
• GSM (Global Mobile System), European standard
• CDPD (Cellular Digital Packet Data), in the USA
  and Canada.
• Ranges of l 100m
• BlueTooth
• Infra-red
• HomeRF.

22
Spontaneous Networking

• The term best describes the integration of mobile
  devices into a given network.
• Encompasses applications that involve connection
  of mobile and non-mobile devices to networks.
• Challenge enable universal interoperability
  between mobile devices and local non-mobile
  services
• e.g., laptops or palmpilots need to detect and be
  able to use available resources, like printers,
  fax machines, etc., when they move into different
  surroundings.

23
Spontaneous Networking (contd)

• Requirements
• Easy connection to a local network
• Avoid the need of pre-installed cabling,
  inconvenience of plugs and sockets
• Transparently reconfigure a mobile device to
  obtain connectivity (avoid the need of manually
  installing drivers).
• Easy integration with local services
• Automatic discovery of available services.
• Active research area.
• Challenge for IP addressing
• Classical IP addressing and routing assumes that
  computers are located on a particular subnetwork
• if a computer is moved to another subnet it is no
  longer accessible with its IP address
• Solution MobileIP (discussed later)

24
Spontaneous Networking (contd)

• Limited connectivity
• Users are intermittently disconnected as they
  move
• Could be disconnected for long periods of time
• Security and Privacy
• Security attacks by mobile devices onto the host
  network or vice versa
• Tracking of physical location of the user
• Access to data otherwise protected by a firewall
• Many other scenarios.

25
Discovery Services

• Accept and store details of services that become
  available on the network and respond to queries
  from clients about them.
• Offer two interfaces
• A registration service accepts registration
  requests from servers and records the details in
  the discovery services database
• A lookup service accepts and processes queries
  concerning available services returns enough
  details to the client to enable it to choose
  among similar services and establish a
  connection.
• Example Jini (discussed later in class).

26
Security Requirements

• Confidentiality
• protection against disclosure to unauthorized
  individuals.
• Integrity
• protection against alteration or corruption.
• Availability
• protection against interference with the means to
  access the resource (denial of service attack).

27
Firewalls

• Creates a protection boundary between the
  organization's intranet and the Internet.
• Runs on a gateway - a computer that stands at the
  network entry point to the intranet.
• Receives and filters all the incoming and
  outgoing messages according to the organizations
  security policy.

28
Secure Network Environment

• Need to move beyond the restrictions imposed by
  firewalls.
• Need to ensure authentication, privacy and
  security over unprotected channels.
• Use of cryptographic techniques.
• Virtual Private Network (VPN) concept
• Use encryption schemes to establish secure
  tunnels through the Internet.

29
Time and Data Delivery

• Most of the data can be delivered within a range
  of transfer rates
• E.g., e-mail, file transfer.
• Time-critical data streams of data that are
  required to be transferred at a certain rate.
• Multimedia data require guaranteed bandwidth and
  bounded latency for the communication channels
  they use.

30
Quality of Service

• The ability to meet deadlines when transmitting
  and processing streams of real-time multimedia
  data
• provide computing and communication resources.
• Currently network performance deteriorates fast
  with load growth
• no QoS support on the Internet.

31
Types of Networks

• Local area networks (LANs).
• Wide area networks (WANs).
• Metropolitan area networks (MANs).
• Wireless networks.
• Internetworks.

32
LANs

• A collection of hosts connected by a high speed
  network of a single communication medium
• twisted pair, coaxial cable, optical fibre.
• Designed and developed for communications and
  resource sharing in a local work environment
• room, campus, building.

33
LANs (cont'd)

• A segment is a section of a cable serving a floor
  or a building
• no routing of messages is required since the
  medium provides direct connection between all of
  the nodes connected to it.
• Larger LANs consist of several segments.
• For a LAN, total system bandwidth is high and
  latency is low.

34
LAN Technologies

• Ethernet as a dominant technology for wired LANs
• lacks latency and bandwidth guarantees needed by
  multimedia applications.
• ATM networks were developed to fill the gap
• their high cost inhibited their adoption for
  LANs.
• High-speed Ethernet
• is deployed in a switched mode
• overcomes drawbacks of Ethernet
• not as effective as ATM for MM data.

35
WANs

• Networks connecting remote communicating
  entities
• lower speed between nodes
• used to connect LANs.
• The communication medium is a set of
  communication circuits linking a set of routers-
  dedicated computers that
• manage the communication network
• rout messages or packets to their destinations.

36
WANs (cont'd)

• Routing operations introduce a delay at each
  point of routing
• total latency for a transmission depends on the
  route taken and traffic encountered.
• Lower bound on latency is set by physical
  properties of the medium
• the speed of electronic signals in most media is
  close to the speed of light.

37
MANs

• Network based on the high-bandwidth copper and
  fibre optic cabling
• installed in metropolitan areas for transmission
  of video, voice, or other multimedia data over
  distances up to 50km.
• Likely to meet requirements set for LANs while
  connecting more distant entities.
• Last mile technology.

38
MAN Technologies

• DSL (digital subscriber line)
• typically uses ATM switches located in telephone
  exchange to route digital data onto twisted pair
• limited range 1.5km from the switch
• speed 0.25-6.0Mbps.
• Cable Modem
• uses analog signalling over coaxial cable
• greater range than DSL
• speed 1.5Mbps.

39
Wireless networks

• Digital wireless communication technologies
• WaveLAN (IEEE 802.11)
• 2-11Mbps over 150m
• wireless local area network designed to replace
  wired LANs.
• other technologies to connect mobile devices to
  other mobile or fixed devices in the immediate
  vicinity.

40
WPANs

• Wireless personal area networks
• infra-red links
• included in laptops and palmtops.
• BlueTooth low-power radio network
  (www.bluetooth.com)
• 1-2 Mbps over 10 m.

41
Mobile phone networks

• Based on digital wireless network technologies.
• Standards
• GSM (global System for Mobile communications)
  used in Europe
• Most mobile phones in the US are based on the
  analog AMPS cellular radio network with CDPD
  (Cellular Digital Packet Data) layer over it.
• Offer wide-area mobile connections to the
  Internet for portable devices
• low-data rates 9.6-19.2 kbps
• successor networks are being designed for
  128-384kbps over km and 2Mbps for smaller cells.

42
Internetworks

• A communication subsystem in which several
  networks are linked together to provide common
  data communication facilities that conceal the
  technologies and protocols of the individual
  component networks and the methods used for their
  interconnection.
• Built upon a variety of LAN and WAN technologies
• interconnected by routers (dedicated switching
  computers) and gateways (general-purpose
  computers)
• a software layer supports addressing and data
  transmission.
• Example the Internet.

43
Network Principles

• Packet transmission.
• Data streaming.
• Switching schemes.
• Protocols.
• Routing.
• Congestion control.
• Internetworking.

44
Packet transmission

• Message sequence of data items (binary).
• Messages are subdivided into packets of bounded
  size
• to manage the buffer storage
• to avoid long wait for a window of sufficient
  size on the communications channel.

45
Data Streaming

• Packet transmission is inappropriate for
  multimedia.
• MM applications rely on the transmission of data
  stream at guaranteed rates with bounded latencies
• QoS requirements
• bandwidth, latency, reliability
• availability of a channel from the source to the
  destination
• buffering where appropriate to cushion flow
  irregularities.

46
Data Streaming (cont'd)

• ATM networks are designed to provide the
  necessary QoS for MM data.
• IPv6 includes feature for recognition and special
  treatment of MM data packets.

47
Switching Schemes

• Broadcast
• no switching everything is transmitted to every
  node
• Broadcast-based technologies
• Ethernet
• Wireless.
• Circuit switching
• a channel is created from the source to the
  destination
• telephone networks are based on circuit
  switching
• referred to as POST (plain old telephone system).

48
Switching Schemes (cont'd)

• Packet switching, or store-and-forward
• no direct channel between the source and the
  destination
• packets are forwarded from node to node along the
  route and buffered if necessary.
• Frame relay
• switch very small packets (frames)
• switching nodes base their decisions on the first
  few bits of the packet
• frames are not stored at nodes but streamed
  through them
• basis for ATM technology.

49
Protocols

• Communication protocol a set of rules and
  formats it defines a specification of
• the sequence of messages exchanged
• the format of the data in the messages.
• Existence of open protocols enables
  component-based software development.
• A protocol is implemented as a pair of software
  modules on the sender and receiver nodes.
• Examples transport protocol (implements
  process-to-process channel) network protocol
  (handles routing).

50
Protocol Layers

• Network softwarehierarchy of layers.
• Each layer provides a service to the layer above
  it and utilizes the services of the layer below.
• Each layer appears to communicate directly to its
  peer on the other side of the network.
• Each layer communicates via local procedure calls
  to the adjacent layers

Layer n
Layer 2
Layer1
51
Data Encapsulation

• Peer protocol modules must communicate control
  information to each other
• e.g., instructions on how to handle the message
  upon arrival, etc.
• A special data structure is attached at either
  end of the message - a header or a tailer.
• The rest of the message is called a body
• info carried over from the layer above.
• Data is encapsulated by a module.

52
Protocol Suits/Stacks

• A complete set of protocol layers.
• Examples OSI (open system interconnection),
  Internet protocol suit.
• Protocol layering
• simplifies and generalizes the software
  interfaces for access to the communication
  services of the networks
• induces performance cost
• N layersN control transfers
• header/tailer data overhead.
• actual transfer rates ltlt available network
  bandwidth!

53
OSI Model
Application
Application
Presentation
Presentation
Session
Session
Transport
Transport
Network
Network
Network
Data link
Data link
Data link
Physical
Physical
Physical
54
Physical Layer

• The physical layer defines electrical signalling
  on the transmission channel how bits are
  converted into electrical current, light pulses
  or any other physical form.
• Specific functions
• connection establishment and termination
• encoding and transmission of bits
• Repeating or amplification to increase the range
  of transmission.

55
Data Link Layer

• Defines how the network layer packets are
  transmitted as bits.
• Examples of data link layer protocols
• PPP (Point to Point Protocol)
• Ethernet framing protocol.
• Bridges work at this layer only.
• Other functions
• Framing and Error detection
• transmission might get corrupted, bits may be
  lost (parity, checksum)
• may lose connection.
• Flow control
• may send data too fast for a modem
• data might get delayed a long time in the network.

56
The Network Layer

• Delivers packets from sending computer to
  receiving computer (host-to-host).
• Defines how information from the transport layer
  is sent over networks and how different hosts are
  addressed.
• Example of a network layer protocol the Internet
  Protocol.
• Device that takes care of the network level
  functions is router or sometimes a gateway .
• Functions
• Addressing Determines which machine to send the
  packet to
• Routing Determines the best set of links
• Congestion Control Routes the packets via a
  different route if one intermediate node gets
  flooded with packets.

57
The Transport Layer

• Takes care of data transfer, ensuring the
  integrity of data if desired by the upper layers.
• Provides end-to-end delivery.
• Functions
• establishing and terminating connection
• flow control
• error detection and correction
• Multiplexing.
• TCP and UDP operate at this layer.

58
The Session Layer

• Establishes and terminates connections and
  arranges sessions to logical parts.
• Provides a means of controlling the dialogue
  between two end users
• Dialogue management (half versus full duplex)
• Synchronization and recovery management.
• This layer is not often used in existing systems.
  
• TCP and RPC provide some functions at this layer.

59
The Presentation Layer

• Takes care of data type conversion
• An example of protocol residing at this layer
  XDR (External Data Representation), which is used
  by RPC applications to provide interoperability
  between heterogeneous computer systems
• Presentation layer functions are, in most
  systems, handled elsewhere in the network
  protocols

60
The Application Layer

• Defines the protocols to be used between the
  application programs.
• Examples of protocols at this layer are
  protocols for electronic mail (e.g. SMTP), file
  transfer (e.g. FTP) and remote login, directory
  look up, http.

61
The Internet Model

• The implementation of the Internet does not
  follow the OSI model.
• Also called TCP/IP model.
• Evolved from ARPANET.
• Note the components are not strictly layered.

Application
Application
TCP/UDP
TCP/UDP
IP
IP
Network
Network
62
The Internet Model (cont'd)

• Network layer
• a combination of hardware (network adapter, etc.)
  and software (network device driver).
• Internet Protocol layer
• creates a logical network over multiple
  networking technologies.
• Transmission Control Protocol and User Datagram
  Protocol layer
• alternative logical channels to application
  programs.
• Application layer
• a set of application protocols to enable
  interoperability of popular applications.

63
The Internet Model (cont'd)

• Does not imply strict layering
• programs are free to define new channel
  abstraction or applications that run on top of
  any of the existing protocols.
• IP as a focal point of the model
• a variety of protocols above IP level and a
  number of implementations under it.

64
Packet Assembly

• Function of the transport protocol.
• Divides messages into packets and assigns
  sequence numbers to them before transmission.
• Reassembles them after transmission according to
  the sequence numbers.
• Encapsulation headerbody(data field).
• Length(body) ltMTU
• maximum transfer unit.
• Ethernet MTU1500 bytes, IP MTU64kbytes.

65
Ports

• Software-definable destination points for
  communication within a host.
• Attached to processes for interprocess
  communications.
• Transport layer obtains a message at a port and
  delivers it to another port
• port numbers are part of the header
  transport addressnetwork addressport number

66
Addressing

• A network address is a unique numeric identifier
  of a host.
• Used by routers to forward frames.
• For the Internet model IP address.

67
Packet Delivery

• Datagram packet delivery (connectionless
  approach).
• A message is not sent as a single unit, but
  broken down into small packets that are
  transmitted individually
• Every packet contains the full network address of
  the source and the destination
• enough information for any switch encountered en
  route to decide how to route the packet
• no circuit set-up is required
• the network retains no info about the packet
• packets may travel on different routes and may
  even arrive to the destination out of order
• delivery is not necessarily affected by failure
  of one or several links.

68
Packet Delivery (cont'd)

• Virtual circuit packet delivery
  (connection-oriented approach).
• virtual connection from the source to the
  destination must be established (dynamically)
• receives a virtual circuit identifier (VCI) to be
  used by the datagrams between the source and the
  destination
• each node maintains a table indicating which link
  should be used for each VC
• no addresses are required and the overhead
  (caused by VCI) is small
• several virtual circuits may use the same link at
  a time
• the connection is broken if a single link fails.
• Example ATM uses this technology.

69
Routing

• Adaptive routing
• the best way for communication is re-evaluated
  periodically
• routing decisions are made on hop-by-hop basis
  using locally held information.
• A routing algorithm
• makes decisions about the rout taken by each
  packet
• in circuit-switched networks all decisions are
  made when the connection is being established
• in packet-switched networks the route is
  determined independently for each packet
• updates its knowledge of the network
• traffic intensity, failed links, etc.

70
Routing (cont'd)

• A problem of graph theory
• networks are representable by graphs.
• Bellman's shortest path algorithm 1957.
• A routing algorithm must be distributed
• centralization is the enemy of scalability.
• Extension to a distributed algorithm by Ford
  Fulkerson 1962
• Bellman-Ford protocols.

71
Routing table update

• Distance vector algorithm
• implemented in RIP (one of the Internet
  protocols)
• each node maintenance/updates a vector of
  "distances" (costs) for each destination on the
  network.
• Link state algorithm
• implemented in OSPF (one of the Internet
  protocols)
• every node maintains and disseminates information
  on how costly it is to reach its immediate
  neighbours
• each node updates its knowledge based on the
  information received from its neighbours
• each node eventually builds a map of the whole
  network.

72
Congestion Control

• Network capacity is limited by the performance of
  its communication links and switching nodes.
• Queues are built at the hosts as the load
  approaches capacity.
• Packets are dropped when buffers are full.
• Dropped packets need to be resent.
• Throughput deteriorates.

73
Congestion Control (cont'd)

• Solution increase delays, keep throughput at its
  maximum
• inform nodes along the route about the state of
  links and switches along the route
• reduce the transmission rate on the route
• buffer packets at the nodes encountered earlier
  on the route.
• Congestion control is achieved by informing nodes
  along a route that congestion has occurred.

74
Congestion Control (cont'd)

• Congestion information is supplied by
• Transmission of choke packets special messages
  requesting a reduction in the transmission rate
• Special provisions in a transmission control
  protocol, e.g., TCP
• Observing occurrence of dropped messages.
• In virtual circuit-based networks congestion
  information is received and acted on at each
  node
• QoS management.

75
Internetworking

• Internetworkintegrated network
• Encompasses many subnets implemented over a
  number of technologies, like Ethernet, ATM, IDSN
  links and DSL connections.
• Requirements
• A unified internetwork addressing scheme
• A protocol defining the packet format and packet
  handling rules
• Internetworking components (hardware) to route
  packets to their destination.
• For the Internet, I and II are provided by IP
  addresses III is performed by Internet Routers.

76
Subnets

• A portion of network that shares a common address
  component.
• On TCP/IP networks, subnets are all devices whose
  IP addresses have the same prefix.
• Networks are divided using a subnet mask
  (discussed later).
• Subnetting facilitates security and performance.

77
Interconnection Devices

• Router a general-purpose computer responsible
  for
• forwarding the internetwork packets that arrive
  on any connection to the correct outgoing
  connection
• Maintaining routing tables for the above purpose.
• Note routing is not required for Ethernet,
  wireless, and other networks where hosts are
  connected to a single transmission medium.
• Bridge a link between networks of different
  types.
• Bridge/Router links several networks and,
  therefore, perform routing.

78
Interconnection Devices (contd)

• Hub a simple connection for hosts on a broadcast
  network
• Provides means of connecting additional hosts
• Overcomes distance limitations (amplifies).
• Switch a router for a local network
• Interconnects several separate Ethernets by
  routing incoming packets onto an appropriate
  network connection
• Starts with no knowledge of the wider
  internetwork and builds up routing tables by
  observation of traffic and supplemental broadcast
  requests.
• Switches vs hubs the former reduce congestion by
  transmitting only to an appropriate network
  connection.

79
Tunnelling

• Hiding of the underlying network protocol.
• Necessary when a pair of nodes need to
  communicate over an alien protocol.
• They construct a protocol tunnel or encapsulate
  the datagram.

Encapsulators
A
B
IP network
IP network
ATM
80
Tunnelling (contd)

• A protocol tunnel is a software layer that
  transmits packets through an alien network
  environment.
• Examples
• MobileIP transmits IP packets to mobile hosts
  anywhere on the Internet by constructing a tunnel
  from their host base.
• PPP protocol for dial-up line constructs a tunnel
  to transmit IP packets.
• ATM Adaptation layer constructs a tunnel to
  transmit IP packets.
• With the anticipated transition from IPv4
  (current version to IPv6, IPv4 will constructs a
  tunnel to transmit IPv6 packets.