Computer Networks

What is the internet?

The internet is a collection of networks with billions of connected computing devices.

• Devices
  • Hosts run network apps.
• Packet switches
  • Forward packets (chunks of data) to their destination.
  • Done by routers and link-layer switches.
• Communication links
  • Fiber, copper, radio spectrum (WiFi, 4G, satelite)
  • Transmission rate: bits per second (bandwith)

Network of networks

• Home networks, enterprise networks, mobile networks
• Internet Service Provider (ISP)
  • hosts connect to the internet via access ISPs
  • most ISPs own part of the physical network (fiber, cables, etc)
• Interconnected ISPs
  • Regional, National, Global ISPs
  • resulting network is very complex
  • evolution driven by economics, national policies

Protocols and standards

A protocol defines the format and the order of messages exchanged between two or more communicating entities, as well as the actions taken on the transmission and/or receipt of a message or other event

• TCP, IP, HTTP etc.
• Internet Standards
  • Request for comments (RFC)
  • IETF: Internet Engineering Task Force

How ISPs are connected

Access networks connect to global ISPs. Global ISPs are connected either by Interent exchange points or by peering.

Peering:

• two or more ISPs interconnect with each other to exchange internet tracfic for mutual benefit.
• public peering (at IXP) or private peering (direct interconnects)

Content providers networks

Big content delivery networks CDNs may run their own networks for better control and service.

Operation principles:

Packet Switching:

Network forwards packets from one router to the next across links on path from source to destination.

Packets:

• data from application is broken into smaller chunks, packets, of length L bits.
• Length is typically is variable but upper bounded.
• A packet contains data and a header.
• The header contains control information e.g. the destination of the packet.

store and forward:

Entire packet must arrive at router before it can be transmitted to the next link.

null

• packet transmission delay: takes L/R seconds to transmit (push out) L-bit packet into link at R bps

packet transmission delay = time needed to transmit L-bit packet into link = L (bits) / R (bits/sec)

Queueing

null

If the arrival rate (in bps) to a link exceeds the transmission rate (bps) of link for some period of time:

• packets will queue, waiting to be transmitted on output link
• packets can be dropped (lost) if memory (buffer) in router fills up

Circuit switching - Alternatice to packet switching

end-end resources allocated to, reserved for “call” between source and destination

dedicated resources: no sharing

• circuit-like (guaranteed) performance
• circuit segment idle if not used by call (no sharing)

FDM

Frequency Division Multiplexing

• optical, electromagnetic frequencies divided into (narrow) frequency bands
• each call allocated its own band, can transmit at max rate of that narrow band

TDM

Time Division Multiplexing

• time divided into slots
• each call allocated periodic slot(s), can transmit at maximum rate of (wider) frequency band (only) during its time slot(s)

Packets vs. Circuits

Two methods for organizing communication in a network:

1. circuit-switching:
   • for each connection between two endpoints a part of the link speed is reserved
   • very suitable if the users need a constant number of bits per second; e.g telephony
   • circuit establishment overhead
2. packet-switching:
   • information flows through the network as packets, which one after the other are sent at the full link speed
   • very suitable if the users’ needs are very variable; e.g., Internet traffic

Layering

Networked systems often use a layered design Because, layering provides conceptual and structural advantages

• a structured way to discuss system components
• modularity makes it easier to update system components Each layer uses the service from the layer below it, to provide a better, more useful, more complete service to the layer above it
• Each layer may introduce its own header “One layer’s header is (part of) another layer’s data.”

Layer	Responsibility	Example	Message Name	Addressing
Application Layer	User applications such as web browsing, email	HTTP, SMTP, etc	Message
Transport Layer	End to end delivery of segments	TCP (reliable, connection-oriented, flow control), UDP (unreliable, connection-less, no flow control)	Segment	Port #
Network Layer	Routing of packets from a source to destination	IP	Datagram	IP address
Link Layer	Unreliable delivery of frames to a neighbor node	WiFi, Ethernet	Frame	MAC address
Physical Layer	Unreliable delivery of bits	wireless, wired (optical links, coaxial cable, copper)

Hourglass Model

Thin waist of the Internet is IP:

• Multiple application, transport protocols above IP
• Multiple link layer and physical layer below IP

Addressing in the Internet

IP address: Unique address showing the recipient

Computers on the Internet are identified by IP addresses:

• a 32-bit number, for example: 10000010 01011001 00000011 11111001
• For convenience, octets are usually written down in decimal, for example 130.89.3.249.

IP addresses are assigned systematically;

• For example, all addresses starting with 130.89 are on the University of Twente (IP = Network ID + Host ID).

For human convenience, many hosts also have a name, for example

• www.utwente.nl.

• Converting names to IP addresses and vice versa is done by the Domain Name System (DNS) 47

• Each sub-network has a subset of the UT's full address range.

• Routers outside the UT only need 1 entry in their forwarding tables for the entire UT; subnetworks are handled within the UT.

IPv4

32-bit addresses 2^32 ~ 4 Billion

But, shortage of IP addresses.

Domains:

• For human convenience, many hosts also have a name, for example www.utwente.nl
• Converting names to IP addresses and vice versa is done by the Domain Name System (DNS)

Routing and Forwarding

Routing: Finding a path between a source and a destination over the most appropriate routers Forwarding: local action, moving incoming packet to appropriate outgoing link of the router

Performance metrics: loss, delay, throughput

How do packet delay and loss occur?

• packet queue in router buffers, waiting for turn for transmission.
  • queue length grows when arrival rate to link (temporarily) exceeds output link capacity
• packet loss occurs when memory for the queue if full.

Best-effort service:

• Internet provides best effort service: no guarantees
• lost packet may be retransmitted by previous node, by source end system, or not at all

packet delay: four sources:

• $d_{nodal}$: total delay at a node e.g, router
• $d_{proc}$: processing delay
  • check bit errors
  • determine output link
  • typically < microseconds
• $d_{queue}$: queue delay
  • time waiting at output link for transmission
  • depends on congestion level of the router.
• $d_{trans}$: transmission delay
  • L: packet length (bits)
  • R: link transmission rate (bps)
  • $d_{trans}=\frac{L}{R}$
• $d_{prop}$: propagation delay
  • d: length of physical link
  • s: propagation speed (~$2\ast {10}^8$ m/sec)
  • $d_{prop}=\frac{d}{s}$

$d_{nodal}=d_{proc}+d_{queue}+d_{trans}+d_{prop}$

Delays for multiple packets:

When considering the delay of multiple packets:

• Some delays add up: e.g., next packet can only be transmitted after previous packet's transmission delay has ended.
• Others don't: e.g., several packets can be propagating on the line simultaneously.

Throughput

Throughput is the rate (bits/time) unit at which bits are being sent from sender to receiver

The average end to end throughput is the bottleneck link: the link with the lowest capacity.

TCP

Connection-oriented protocol

• A logical connection is established between the two ends before data transfer
• Connection establishment via three-way handshake
• Bidirectional connection, data flow in two directions
• Reliable delivery
• reliable: every transmission of data is acknowledged by the receiver.
• packet loss and corruption might happen but TCP finds it out.
• retransmission of lost packets:
• If the sender does not receive acknowledgement within a specified amount of time, the sender retransmits the data
• Flow control, congestion control

TCP segment:

Source and destination ports (distinguish between multiple connections) • Sequence number (used to order packets) • the first byte of data included in the segment • Acknowledgement number (used to verify packets are received) • indicates the byte number of the next data that is expected to be received • All bytes up through this number have already been received

Connection establishment

Threee way handshake:

1. Host_A: SYN: I want to connect and I will send my data starting with sequence number Initial Sequence Number (ISN) X.
2. Server_B: SYN, ACK: Yes you can connect, I will send my data starting with sequence number (ISN) Y, and expect X+1 from you,
3. Host_A: ACK: Sounds good, I expect Y+1 from you.

null

Problems:

What TCP does in case:

• Channel results in bit errors and corruption of the message.
  1. Receiver receives all data but detects errors
  2. Receiver tells the sender: “please repeat that”
  3. Sender retransmits the data that were corrupted
• Packet loss and corruption.
  1. Receiver receives some data, some is lost, and some is corrupted
  2. Receiver detects errors but cannot always detect loss
  3. Sender must wait for acknowledgment (“ACK” or “OK”)
  4. and retransmit data after some time if no ACK arrives

TCP header fields:

• Seq is the sequence number of the first byte of application-layer data in this packet; or, if this packet does not contain any application-layer data, it is the sequence number that such a first byte "would have" had if it were there.
• Ack is the sequence number of the next byte expected by the sender of this packet; it acknowledges all bytes with lower sequence numbers (therefore it is called a "cumulative" acknowledgement).
• Len is the number of application-layer bytes in this packet (only in Wireshark).
• SYN begin connection.
• FIN end connection.

User Datagram Protocol (UDP):

• Conectionless
• Unreliable
• Packets may or may not arrive at destination
• no handshaking before sending a segment
• best-effort approach

UDP pros:

• small overhead, no need to establish a connection.
• no need to keep a state at sender and receiver.
• video streaming used UDP, nowadays it uses TCP/HTTP video streaming.