Exam Study Notes Flashcards
Explain NAT
Network Address Translation.
Motivation: Local network uses just one IP address as far as the outside world is concerned.
Works: All leaving packets have the same source IP (the IP of the router). The router maps the original source IP and port to the NAT IP and a newly created port. Based on this port all incoming traffic can be directed to the correct devices within the network.
Allows: us to cut down on IP addresses, change internal local IP addresses without affecting the world, inside devices can’t be directly accessed from outside world.
Coaxial Cable Attributes
- Two copper conductors.
- Bidirectional
- Broadband (multiple channels on cable)
Radio Attributes
- Bidirectional
2. Environment effects the signal
Forms of Radio Transmission
- Terrestrial Microwave
- LAN (Wi-Fi)
- Wide-area network (Cellular).
- Satellite
Circuit Switching Attributes
- Dedicated Resources and Guaranteed performance
- Idle if not used.
- Cannot add more users than the defined limit, once resources have been allocated.
Packet Switching Attributes
- Allows more users than Circuit Switching
- Simpler than Circuit Switching
- Excessive congestion possible
Four sources of delay
- Propagation
- Transmission
- Nodal Processing
- Queuing
Propagation Delay Attributes
- Nothing can be done about this form of delay
2. Occurs because of the physical distance between the two communicating nodes.
Transmission Delay Attributes
Key idea: all about the packet size (in bits) and the transmission rate (number of bits per second) the link can deliver.
Nodal Processing Delay Attributes
- Check bit errors
2. Determining output link at each hop
Queuing Delay Attributes
- Represents the time spent waiting at output link for transmission
- Depends on congestion level of router
What is Traceroute and how does it work
Provides network analysis information, E.g. measuring network congestion and delay, etc.
Achieves this by iteratively sending (3) three packets to each of the hosts along the path to the destination and measuring response times.
Where does throughput bottlenecking usually occur?
Client side
Internet Protocol Stack Layers
- Application
- Transport
- Network
- Link
- Physical
ISO/OSI Model
ISO spelt backwards is OSI
- Application
- Presentation (allows applications to interpret data)
- Session (recovery and syncing of data)
- Transport
- Network
- Link
- Physical
TCP and UDP Socket Attributes
TCP sockets are based on a connection
UDP sockets are connectionless
TCP sockets do a 3-way handshake
UDP just sends the sockets straight through
TCP Socket Identification
When creating a TCP socket, need to specify:
- Source IP
- Source Port Number
- Destination IP
- Source Port Number
UDP Socket Identification
When creating a UDP socket, need to specify:
- Destination IP
- Destination Port Number
Explain Stop and Wait (Reliable Data Transfer UDP)
Sender sends a packet and then waits for acknowledgement for the specified packet.
If no acknowledgement is returned, after waiting a ‘reasonable amount of time’ the sender will retransmit the packet once again.
Performance is poor
How does the Go-Back-N (GBN) Protocol work
Keeps a timer for the oldest transmitted (unacknowledged) bit. Should the timer expire, all bits in the same window size are retransmitted.
Drops all out of order packets, till the next in-order packet is delivered.
No buffer.
TCP Attributes
- Point-to-Point
- Reliable
- Connection based
- Fully duplex data (bidirectional)
- Flow controlled
TCP Reliable Transfer Protocols
TCP Fast Retransmit (detecting packet loss based on duplicate acknowledgements)
AMID
Cookies
Permit websites to learn about you. Can be used for:
- Personalised Recommendations
- Personalised Ads
- Authorisation
DNS Capabilities
- Host name to IP translation
- Host aliasing (conical name vs alias name)
- Mail server aliasing
- Load distribution (many IP addresses may correspond to one ‘name’).
TLD
Top level domain (DNS Server)
These are often cached within the local DNS servers (name servers), so they don’t have to visit the root DNS server that often.
How does DNS work
Has a hierarchical approach:
- Uses Root DNS server to find TLD server
- Uses TLD server to find host IP address
- Uses Authoritative server to find IP address for main webpage.
DNS Caching
Once a website is visited, the name server caches the address directly for future uses. This caching has an expiry link (Time to live (TTL)).
Sometimes cached entries can be out of date (should name host change its IP address). Cannot be updated until TTL expires.
Authoritative DNS servers (name servers)
References the organisation’s own DNS server.
Resource Record
DNS Server Entry. Type = MX (used for Mail Server) Type = A (used for website) Type = CNAME (used for conical name) Format (Name, Value, Type, tt1)
IGP
Intra Gateway Protocol.
Examples:
1. RIP - Routing Information Protocol (Distance Vector)
2. OSPF - Open Shortest Path First (Link State - Dijkstra’s Algorithm)
3. IGRP - Interior Gateway Routing Protocol (Cisco)
Routing Information Protocol (RIP)
Uses Distance Vector Algorithm Hops used for distance estimations Maximum of 15 hops Maximum of 25 devices in subnet Exchanges distance vector information with neighbours every 30 seconds
Open Shortest Path First (OPSF)
Uses link state algorithm (Dijkstra’s)
IS-IS routing protocol is nearly identical.
Advertisements flooded to entire AS
BGP
Border Gateway Protocol (glue that holds the internet together)
eBGP: obtains subnet reachability information
iBGP: distributes the subnet reachability information to the entire AS
Where is link layer implemented
In the network interface card within a host system
Name the different forms of error detection and correction
- Cyclic Redundancy Check (CRC)
- Internet checksum
- Parity checking
How does Cyclic Redundancy Check (CRC) work
Views data as ‘D’ number of bits.
Concatenates ‘R’ bits where (D+R) bits are exactly divisible by G (an R + 1 bits pattern).
Transmit (D+R). If the number if exactly divisible by G (e.g. D+R mod G == 0) no errors have been detected.
Can find up to R errors?
CDMA
Code Division Multiple Access
All users share the same frequency.
Each user transmitting data encodes his/her data with the code allocated to the device, based on which the router can distinguish them.
CSMA
Carrier SENSE multiple access
Key idea: listen (SENSE) before you transmit.
TDMA
Time Division Multiple Access
FDMA
Frequency Division Multiple Access
MAC
Media Access Control
Examples: FDMA, TDMA, CDMA
Multiple Access Protocols
- Random Access
- Carrier Sense Multiple Access (CSMA)
- Taking Turns
Collision Avoidance Technique
Reserve Channel Resources using RTS; Continue transmitting till complete (other network devices do not make requests for a period of time when this is happening).
RTS
Request to Send
Able to continually keep transmitting data till acknowledgement is returned
CTS
Clear to Send
Examples of Taking Turns
Polling: Computer is delegated as master, and invites other computers to send through any data that they may wish to transmit.
Token passing: control token that allows a device to transmit next is passed from device to device.
IP Address Parts
Made up of two parts.
- Subnet part
- Host part
Queuing delay and loss
Occurs at the input and output ports of the router.
1st reason it occurs is because of HOL (Head of line) blocking in the input ports.
2nd reason it occurs may be because the output port forwarding rate may be slower than the switch fabric’s rate of parsing the bits to it. And so the buffer overflows and packets are lost!
How much buffering for input and output ports
RTT times link capacity (C)
IP Fragmentation
Links have a max MTU (Maximum Packet Transmission Size) and if the packet you try to transmit is too large, you need to break into smaller packets before you transmit. They are recombined at the receiver.
Distance Vector
Key idea: from time to time, each node sends its own distance vector estimate to neighbours.
Neighbour nodes then update their own distance vectors based on the Distance Vector algorithm.
2D Parity Checking
Can be used to find and correct a single bit error
