BIM 310: Computer Networks

1
BIM 310 Computer Networks

• Time Wednesday 930-1200,
• Thursdays 1100-1300
• Location BLab4
• Instructor Emre Kaçmaz Alper Bilge
• Grading
• Midterm I 20
• Midterm II 25
• 1 Final 40
• Homeworks - 15

2
High-level picture of the problem
Transmission System
Sender Node
Destination Node
Problem Transmit a message M from a source node
to one or more destination node(s) through a
transmission systemcomputer network
Nodes Things that send/receive messages.
Examples are PCs, labtops, PDAs, Internet
telephones etc.
3
Transmission System
Sender Node
Destination Node

• mesh of interconnected routers (switches)
  inter-connected in an arbitrary topology
• the fundamental question how is data transferred
  through net?
• circuit switching dedicated circuit per call
  telephone net
• packet-switching data sent thru net in discrete
  chunks

4
Circuit-Switching Idea

• End-end resources reserved for call
• call setup required
• After the call, the resources (the circuit
  bandwidth) is dedicated and is not shared with
  other calls
• circuit-like (guaranteed) performance
• This course is not about circuit-switching, but
  we will touch on it so that you get an idea on
  how it works

5
Circuit-Switching Data Flow

• After the call is setup, the data flows through
  the circuit bit by bit
• No store or forward delay at the routers
  (switches)
• As soon as a bit from the connection arrives at a
  router, it is immediately forwarded over the
  outgoing link without any delay
• So the transmission time is independent of the
  of links from the source to the destination

6
CS Sharing Link Capacity

• How do several calls using the same link share
  the link?
• In the example above, we have 2 calls sharing 2
  links in the middle of the network
• 2 Approaches
• Frequency Division Multiplexing (FDM)
• Time Division Multiplexing (TDM)

7
Circuit-Switching FDM

• FDM Divide the link capacity into several
  frequency bands (space-wise division) and
  allocate each band to a different call
• Each circuit gets the fraction of the bandwidth
  continuously (all the time)
• Also used by radio/TV transmission through the
  air

8
Circuit-Switching TDM

• TDM Divide the link capacity in time into
  several slots and allocate each slot to a
  different call
• Each circuit gets ALL of the link bandwidth
  periodically
• Used by wireless telephones (GSM)

9
Circuit Switching Example

• 1890-current Phone network
• Fixed bit rate
• Mostly voice
• Not fault-tolerant
• Components extremely reliable
• Global application-level knowledge throughout
  network

10
Circuit Switching Summary

• Establish a dedicated circuit before sending data
• Dedicated resources
• Data flows through the circuit
• No store and forward at the switches (routers)
• Good for constant-bit-rate traffic such as voice
• BUT
• Dedicated resources means if no data is flowing
  the circuit, the allocated resources are idle
• Leads to waste of network resources
• Might be OK for telephone calls where two parties
  are typically talking all the time
• But is this good if two people are exchanging a
  data as in a instant messaging session?
• How can be let other people use unused bandwidth?
• Packet Switching -- Next

11
Packet-Switching Idea
Message M
queue of packets waiting for output link
Message M

• Divide the message into smaller chunks, packets
• Send each packet through the network
  independently
• Each packet uses a links full bandwidth during
  transmission
• Resources are used as needed

12
Packet switching vs Circuit Switching

• 1 Mbit link
• each user
• 100Kbps when active
• active 10 of time
• circuit-switching
• can admit 10 users
• packet switching
• With 35 users, probability gt 10 active less that
  .004

N users
1 Mbps link

• Conclusion Packet switching allows more users to
  use network!

13
Packet-Switching Link Sharing

• resource contention
• aggregate resource demand can exceed amount
  available
• congestion packets queue, wait for link use
• store and forward packets move one hop at a time
• A router must receive the whole packet before the
  packet can be forwarded
• After reception, queue the packet internally and
  have it wait its turn for the output link.
• This is done by each router Per hop forwarding
• Sequence of A B packets does not have fixed
  pattern
• Called statistical multiplexing.

14
Packet-Switching Issues

• Two Fundamental Questions must be answered in a
  packet-switched network
• What should the packet size be?
• Fixed-size or variable-sized packets?
• How big?
• Should we establish an end-to-end path through
  the network for the packets to flow?
• Yes Virtual-Circuit Networks (X.25, Frame-Relay,
  ATM)
• No Datagram Networks (the Internet)

15
Packet-Switching Packet Size
Destination
Source
R1
R2
0
5
10
15

• Consider a message M, that is 7.5106 bits long,
  no message frag.
• Assume each link has 1.5Mbps bandwidth
• It takes (7.5106 /1.5Mbps) 5 seconds to move
  the message from the source to the first switch
  (router) R1
• Another 5 secs to move M from R1 to R2
• Another 5 secs to move M from R2 to destination
• Total time 5 5 5 15 seconds

16
Packet-Switching Packet Size
Destination D
Source S
R1
R2
0
1
1
2
1
2
3
1
2
3
2
3
5
3
5002
5000

• Assume now that we divide the message into 5000
  packets, each 1500 bits long
• It takes 1 milisecs to move the 1st packet from S
  to R1
• But while the first packet is being moved from R1
  to R2, we are also moving 2nd packet from S to R1
• The first packet makes it to D in time 3ms, the
  2nd packet is at R2 and 3rd packet is at R1 at
  this time
• Following this logic, the last packet makes it to
  D at time 5002 ms 5.002 sec as opposed to 15
  seconds

17
Packet-Switching Packet Size

• Small-sized packets has yet another advantage
• Bit errors can be introduced as packet travels
  through the network. In such cases the packet is
  simply discarded
• The smaller the packet, the smaller the discarded
  info
• If small packets are so good, why not make them 1
  byte
• Each packet carries some headers with it
• Headers are used for packet forwarding, and other
  stuff
• The smaller the packet, the bigger the header to
  payload ratio, which translates to more waste of
  bandwidth
• Consider 100 byte packets with 20 byte header
• 20 bandwidth waste
• Consider 1000 byte packets with 20 byte header
• 2 bandwidth waste

18
Virtual Circuits Networks Signaling
C
R5
R2
R10
R3
A
R6
D
R9
R4
R1
R7
B
R8

• Virtual Circuit Networks (e.g., X.25, Frame
  Relay, ATM)
• Establish a path along which the packets will
  flow between the source and the destination. How?
• Use a signaling (virtual circuit establishment)
  protocol
• Ex B tells its router (R1) that it wants to talk
  to C
• The call establishment message is forwarded by
  the routers in the network until it reaches C.
  Then a reply comes back from C to B.
• Path established at call setup time remains fixed
  during packet exchange
• Routers maintain state information for ongoing
  connections

19
Virtual Circuits Networks Forwarding
C
R5
A
R2
R10
R3
3
3
1
45
R6
43
53
2
69
D
2
22
2
R9
1
9
R4
66
R1
77
12
R7
B
R8
VC table at R1
VC table at R2

• each packet carries tag (virtual circuit ID),
  which determines next hop
• Path established at call setup time remains fixed
  during packet exchange
• Routers maintain state information for ongoing
  connections

20
Datagram Networks Idea
C
R5
A
R2
R10
R3
C
C
C
D
R6
D
D
C
D
C
C
R9
D
R4
C
R1
D
C
C
D
R7
D
B
D
R8
D

• Datagram networks (e.g. the Internet)
• No call establishment before data exchange
• Simply put the destination address on top of the
  packet and submit it to the network for delivery
• Similar to postal service

21
Datagram Networks Forwarding
C
R5
A
R2
R10
R3
C
C
C
3
1
D
R6
D
D
2
C
D
2
C
C
R9
1
D
R4
C
R1
D
C
R7
B
R8
Forwarding table at R1
Forwarding table at R2

• Destination address is written on top of a packet
  and it is simply submitted to the network for
  delivery (like postal service)
• Routers look at destination address in packet to
  determine the next hop
• No connection-state information needed in the
  routers
• Routes may change during session

22
Packet switching versus circuit switching

• Is packet switching a slam dunk winner?

• Great for bursty data
• resource sharing
• But, excessive congestion packet delay and loss
• protocols needed for reliable data transfer,
  congestion control
• Q How to provide circuit-like behavior?
• Bandwidth guarantees needed for audio/video apps
• Active research area IP QoS

23
Network Taxonomy
Telecommunication networks
24
Packet Switched Networks

• This course is about packet-switched networks
• We will not cover circuit-switched networks
• In looking at packet-switched networks, our
  approach will be from the view of network
  designer, a system engineer, who wants to build a
  packet switched network from the ground up
• How do you build a packet switched network?
• What are the issues?
• How do you solve them?
• What are the specific solutions in existence
  today?
• We will mostly look at Internet Protocols

25
Point-to-Point Links
B
A
Simple point-to-point link
Message M

• The simplest packet switched network is a network
  consisting of 2 hosts, A and B, and a link
  connecting them
• Link can be guided media, i.e., a copper, coax,
  fiber wire
• Link can be unguided media, i.e., the air
  wireless
• Link can be half-duplex (only one node can send
  data over the link at any time) or full-duplex (A
  can send a message to B, while B is sending a
  message to A)
• Problem Given a message M at A, divide the
  message into several packets, and send them over
  the link to B

26
Point-to-Point Links
B
A
Simple point-to-point link
Message M

• What are the issues in a point-to-point link?
• How does B know the beginning and end of a
  packet?
• Called the framing problem
• How does B know whether the packet is corrupted,
  i.e., if any bits of the message has changed,
  during transmission or not? If any bits changed,
  can B correct them?
• Called the error detection correction problem
• How do you encode a digital data on the link?
• Called the data encoding problem

27
Broadcast (Multi-Access) Links
B
A
Message M
D
C

• The next-simple packet switched network you can
  imagine is a network consisting of several hosts,
  A, B, C and D above, sharing a common link
• Again, the link can be wired or wireless
• In such a network, when one node sends a packet
  over the link, the packet reaches ALL nodes
  attached to the link
• Such a link is called a broadcast link, e.g.,
  Ethernet, FDDI

28
Broadcast (Multi-Access) Links
B
A
Message M
D
C

• What are the issues in a broadcast link?
• All the issues of a point-to-point exists
  framing, error detection correction and
  encoding.
• What else? First issue is, how do the stations
  agree on who gets to use the link?
• Called the media access control problem
• Second, how does A tell that the packet is
  destined to B not to C or D?
• Addressing problem Each station must have a
  UNIQUE address, called the Media Access Control
  (MAC) address

29
Limit on Broadcast Links
B
A
Message M
D
C

• Whats the limit of a broadcast link?
• How many hosts (stations) can we connect to a
  broadcast link?
• Can we build a global network such as the
  Internet with a broadcast link?
• Can you imagine connecting millions of hosts to a
  broadcast link?
• If we do, does it make sense that when a host in
  Germany wants to send a packet to another host
  next door, that my host in here receive that
  packet, examine it, realize that the packet is
  destined to someone else and discard it?
• Broadcast does not scale. So there is a limit on
  the size of a broadcast link.

30
A General Packet-Switched Network

• To build a global packet-switched network such as
  the Internet, we must have a network core
  consisting of lots of packet switches, called
  routers
• The end systems (hosts, stations) are at the edge
  of the network
• End system hosts can be attached to the network
  core with a point-to-point link or they can be
  attached together with a broadcast link and then
  attached to the network core

31
A General Packet-Switched Network

• What are the issues in such a packet-switched
  network?
• Addressing Each host and router interface must
  have GLOBALLY UNIQUE addresses IP address
• When host A wants to send a packet to host F on
  the other side of the network, how does A and
  routers know how to reach F?
• Routing and forwarding problem Establishing
  reach-ability information (forwarding table) and
  using it to forward a packet from the source to
  the destination host

32
A General View of the Internet
Connection to national ISP
router
workstation
local ISP
server
mobile
regional ISP
a company network
a university network

• Mesh of interconnected autonomous systems

33
What about applications?
Web Server
C
R5
FTP Client
Web Browser
R2
R10
R3
R6
A
D
R9
R4
R1
FTP Server
R7
B
R8

• It is the applications that communicate!
• Host A runs a Web Browser and an FTP Client
• Web Browser is talking to the Web Server running
  C
• FTP Client is talking to the FTP Server running
  in D
• Packets from both C and D arrive at A

34
What about applications?
Web Server
C
R5
Web Browser
FTP Client
R2
R10
R3
R6
A
D
R9
R4
R1
FTP Server
R7
B
R8

• What are the issues here?
• How does host A know that green packets need to
  delivered to the Web Browser and Blue Packets
  need to be delivered to the FTP client?
• Multiplexing/Demultiplexing problem
• What if some of the packets sent from the Web
  Server is lost during transmission. How do we
  recover them?
• Reliable packet delivery problem

35
How do two network entities talk to each other
PROTOCOLS

• For two entities to communicate, they must speak
  the same language
• What is communicated?
• Message format
• How is it communicated and what it means?
• Order of messages and their meaning
• When is it communicated?
• Timing of the messages
• The above must conform to mutually acceptable
  conventions between the entities involved
• In networking, these conventions are referred to
  as a protocol

36
Protocols

• A protocol is a set of rules governing the
  exchange of data between the two entities
• Key elements of a protocol are
• Syntax Message format
• Semantics The meaning of messages
• order of messages sent and received
• actions taken on message transmission, receipt
• Timing Includes speed matching and sequencing
• When to send a message

37
More on Protocols

• a human protocol and a computer network protocol

Hi
TCP connection req
Hi

• What are some other human protocols?
• Raise you hand before asking questions
• Take turns to speak, i.e., do not speak at the
  same time

38
Designing Protocols
B
A
Message M

• Recall the issues in communication over a P-2-P
  link
• Message fragmentation Dividing a message into
  packets at A
• Framing Identifying the beginning and end of a
  packet at B
• Error Detection Correction Identifying corrupt
  packets at B
• Encoding Encoding packet bits onto the link as a
  signal at A and reconstructing the packet bits
  from the received signals at B
• What kind of protocols do we need to handle the
  above issues?
• Next

39
Physical and Link Layers
Message M
A
B
link physical
link physical

• Typically the listed issues are handled by 2
  protocols
• A Physical Layer (PL), which deals with bit
  Encoding/Decoding
• Physical Layer at A deals with the following
  problem
• Given a sequence of bits (bits making up a
  packet), how do you encode the bits onto the link
  as signals (electromagnetic, light..)
• Physical Layer at B deals with the following
  problem
• As you receive signals from the link, how do you
  decode these signals into bits?
• A Link Layer (LL) that sits on top of the
  physical layer (PL) and deals with the rest of
  the problems Message Fragmentation, Framing,
  Error Detection/Recovery

40
Physical and Link Layers

• What about a broadcast link?
• PL and LL will be there as before having the same
  responsibilities as described before
• But now LL has the additional responsibility of
  Media Access Control (MAC) to deal with
• Link Layer data transfer between neighboring
  network elements
• PPP, Ethernet
• Physical Layer bits on the wire

41
Network Layer
link physical
link physical
link physical
link physical
R1
B
R2
A

• What about a general packet-switched network?
• Are PL and LL enough?
• Recall that LL is responsible for data transfer
  between neighboring network elements, that is,
  if they are connected to the same link
• Are hosts A and B neighbors above? No.
• Need a new layer, called the Network Layer
• Responsible for forwarding of datagrams from
  source HOST A to destination HOST B
• Internet Protocol (IP, routing protocols)

42
Transport Layer
Web Browser
Web Server
FTP Client
B
FTP Server
A
R1
R2
C

• What about applications running in hosts?
• Is NL enough?
• Recall that NL is responsible for forwarding a
  packet from one HOST to another HOST
• How do you make applications on HOSTs to
  communicate?
• Need a new layer, called the Transport Layer
• Responsible for providing communication between
  applications running in different hosts
• A Web Browser talking to a Web Server

43
Application Layer
Web Server
FTP Client
Web Browser
B
FTP Server
A
R1
R2
C

• Lots of different applications in the Internet
• Web browsing, file download, e-mail, instant
  messages, presence
• Each require different message types, formats,
  actions
• So need a new layer, called the Application
  Layer
• Responsible for defining application specific
  message types, formats, actions taken on messages
• HTTP for Web, FTP for file download, SMTP for
  e-mail, SIP for instant messaging and presence
  so many others!!

44
Internet protocol stack

• application Define application specific message
  types, formats
• FTP, SMTP, STTP
• transport Provide application-to-application
  communication
• TCP, UDP
• network Provide host-to-host communication. That
  is, forwarding of packets from source to
  destination
• IP, routing protocols
• link Provide data transfer between neighboring
  network elements (host-to-host, host-to-router,
  router-to-router)
• PPP, Ethernet
• physical transmit bits on the link

45
Protocol layering and data

• At the source, each layer takes data from above
• adds header information to create new data unit
• Called encapsulation
• passes new data unit to layer below
• At the destination, each layer takes data from
  below
• strips off its own header
• Called decapsulation
• passes the remaining part of the packet to the
  upper layer

source
destination
message
segment
datagram
frame
46
Multiplexing/Demultiplexing
MUX

• A way for multiple protocol objects at one level
  to identify themselves to the protocol above or
  below them.
• Multiplex
• Tag each message with a key
• Lower protocol knows where it came from!
• Demux
• Use key on arriving packet to know where to send
  it above

1
2
3
data
1
data
3
DEMUX
1
2
3
data
1
data
3
data
1
data
3
47
Protocol Interfaces

• Each protocol defines 2 interfaces
• Service Interface The kind of services it
  provides to protocols that sit on top of it on
  the same machine
• Peer Interface Communication interface with its
  counterpart (peer) on another machine
• This interface defines the form and meaning of
  messages exchanged between protocol peers to
  implement the service interface

• Example IP exports a connectionless, unreliable,
  best-effort datagram service to transport layer
  protocols

48
Protocol Communication

• A protocol always communicates with same protocol
  at peer machine. Never do we have a protocol at
  one layer talk to another protocol at a different
  layer at the peer

49
Internet Protocols

• Defined by Internet Engineering Task Force (IETF)
• Hourglass Design Everything goes over IP
• Lots of application layer protocols
• Mainly 2 transport layer protocols TCP, UDP
• Network Protocol is Internet Protocol (IP)
• Any Link Layer Protocol

80
4444
20,21
6
17
50
OSI v TCP/IP

• Open Systems Interconnection
• Developed by the International Organization for
  Standardization (ISO)
• Seven layers
• A theoretical system delivered too late!
• TCP/IP is the de facto standard

51
Network Performance Metrics

• Bandwidth
• data transmitted per time unit
• link versus end-to-end
• Notation Mbps 106 bits per second
• Bits transmitted at a particular bandwidth can be
  regarded as having some width
• (a) Bits transmitted at 1Mbps, each bit is 1us
  wide
• (b) Bits transmitted at 2Mbps, each bit is 0.5us
  wide

(a)
(b)
52
Network Performance Metrics

• Latency (delay)
• time for the first byte of the message to reach
  the destination
• one-way versus round-trip time (RTT)
• components
• Latency Transmission Propagation Nodal
  Pros. Queuing Delay
• Transmission Time Message Size / Bandwidth
• Propagation Distance / Speed of Light
• Nodal Processing F(Amount of Processing,
  Processor Speed)
• Queuing Delay F(Amount of total traffic)

transmission
A
transmission
C
B
nodal processing
queueing
53
Latency or Delay

• dtrans transmission delay
• L/R, L Message size, R Link Bandwidth
• significant for low-speed links
• dprop propagation delay
• a few microsecs to hundreds of msecs
• dproc processing delay
• typically a few microsecs or less
• dqueue queuing delay
• depends on congestion (the amount of total
  traffic)

54
Queueing delay

• Rlink bandwidth (bps)
• Lpacket length (bits)
• aaverage packet arrival rate

traffic intensity La/R

• La/R 0 average queueing delay small
• La/R -gt 1 delays become large
• La/R gt 1 more work arriving than can be
  serviced, average delay infinite!

55
How and why do packet loss occur?

• Packets get queued in router buffers
• If packet arrival rate exceeds output capacity,
  packets get buffered and wait for their turn to
  be transmitted
• Buffer is of finite size
• If more packets than what buffer can store, new
  packets will be dropped

A
B
56
Introduction Summary

• Covered a ton of material!
• Internet overview
• whats a protocol?
• network edge, core
• packet-switching versus circuit-switching
• performance loss, delay
• layering and service models

• You now have
• context, overview, feel of networking
• The rest of the course will be to learn the
  details of these protocols in the context of IP
  protocol stack