CS331 · Computer Networks

Advantages & disadvantages, every chapter

Every advantage and disadvantage the slides give, one tab per chapter and one card per topic. Cards carry a note when the topic was already discussed in an earlier chapter, and when it comes back later.

Advantages

Disadvantages

Already discussed earlier

Comes back later

OVERVIEW

Topics that come up in more than one chapter

Each row is a topic that builds on something an earlier chapter already discussed. The chapter tabs mark these with an Already discussed note, and mark the first appearance with Comes back.

Where	Topic	Already discussed in
Ch 02	Amplifier vs repeater (fixing attenuation)	Ch 1 Part 2: the repeater is listed as a physical-layer device.
Ch 03	Controlled access (reservation, polling, token)	Ch 1: token passing circulates around a ring topology.
Ch 04	Traditional (half-duplex) Ethernet	Ch 3: CSMA/CD — its pros and cons are there. Ch 1: bus and star topologies.
Ch 04	Full-duplex (switched) Ethernet	Ch 1: full duplex as a transmission mode (telephone).
Ch 04	Faster Ethernet standards	Ch 2: fibre's advantages and costs.
Ch 06	Hub (physical layer)	Ch 1: the hub is the centre of a star. Ch 2: a hub works like a multi-port repeater.
Ch 06	Link-layer switch	Ch 4: full-duplex switched Ethernet doubles throughput.
Ch 08	Network-specific and default routing	Ch 7: the default gateway on every host.
Ch 09	TCP	Ch 7: virtual-circuit (connection-oriented) switching. Ch 3: retransmission-based error control.
Ch 09	UDP	Ch 7: datagram (connectionless) switching. Ch 3: why retransmission is bad for real-time traffic.
Ch 09	Connectionless vs connection-oriented service	Ch 7: the same split at the network layer (virtual circuit vs datagram).
Ch 10	DNS over UDP	Ch 9: UDP's advantages and disadvantages.

How to use this sheet

In a scenario answer

Name the choice, define it, then pick the advantages that match the scenario’s words and give at least one honest disadvantage.

In a comparison question

Pairs that are designed as opposites — UTP vs fibre, hub vs switch, store-and-forward vs cut-through, distance vector vs link state, TCP vs UDP — mirror each other: one side’s advantage is the other’s disadvantage.

CHAPTER 01

Introduction

Chapter 1 · Parts 1 and 2: networks, topologies, network types, standards, layering. 9 topics.

1.1

Mesh topology

Every device has a dedicated point-to-point link to every other device; N(N − 1)/2 links.

Advantages

Dedicated links: secure, no traffic problems (no congestion)
Robust: if one link fails the network still works
Easy to discover faulty links

Disadvantages

Difficult to install and update (too much wiring)
Expensive: many cables, and N − 1 I/O ports (NICs) per device

Source: Ch 1 Part 1 slides

1.2

Star topology

Each device has a dedicated link to one central controller (hub or switch); N links.

Advantages

Less expensive than mesh
Easy to install
Easy to add and remove devices
Robust: one broken link affects one device
Easy to discover faulty links
Secure, but less than mesh

Disadvantages

If the central device fails, the whole network stops (single point of failure)

Comes backCh 4: Ethernet runs as a star. Ch 6: whether the centre is a hub or a switch changes the collision domains.

Source: Ch 1 Part 1 slides

1.3

Bus topology

One main backbone cable; each device taps in with a drop line; 1 + N links.

Advantages

Easy to install
Less cabling than mesh and star

Disadvantages

Difficult to isolate a fault
Difficult to add new devices
A break in the bus cable stops all transmission

Comes backCh 3 and Ch 4: a shared bus needs a media access method (CSMA/CD).

Source: Ch 1 Part 1 slides

1.4

Ring topology

Each device links to the two devices on either side of it; N links.

Advantages

Easy to install
Less cable
Easy to add or remove a device
Fault isolation is easy

Disadvantages

Not robust: one break can stop the ring

Comes backCh 3: token passing is the access method designed for a ring.

Source: Ch 1 Part 1 slides

1.5

Hybrid topology

A combination, e.g. a star backbone joining bus networks (star–bus, star–ring).

Advantages

Combines the strengths of the topologies used
Lets existing networks be joined centrally

Disadvantages

Inherits the weaknesses of its parts (a bus break, a central-device failure)

Source: Ch 1 Part 1 slides (figure); pros/cons follow from the parts

1.6

Throughput vs delay (network criteria)

Performance is measured by throughput and delay.

Advantages

More throughput lets users send more data

Disadvantages

Throughput and delay contradict: more data in the network raises the delay

Source: Ch 1 Part 1 slides

1.7

Protocols and standards

Agreed rules so equipment from different vendors can talk.

Advantages

Ensure a large market for equipment and software
Give guidelines to manufacturers, vendors and government agencies
Allow products from different vendors to communicate (interoperability)

Disadvantages

Without them, each vendor's network needs major translation software to talk to others (the 1970s problem)

Source: Ch 1 Part 1 slides

1.8

Layered architecture

Network functions grouped into layers; layer n − 1 serves layer n through an interface.

Advantages

Breaks a complicated process into simpler ones: easier hardware and software implementation
Each layer is implemented independently, so different manufacturers can supply different layers (collaboration and competition)
A layer's implementation can be replaced without affecting the others

Disadvantages

Complexity of a device increases with the number of layers it supports

Source: Ch 1 Part 2 slides

1.9

Peer-to-peer vs client/server

From Tutorial 1: who provides the services.

Advantages

Peer-to-peer: no central server, so no single point of failure; cheap; each machine is client and server
Client/server: central management, backups and security

Disadvantages

Peer-to-peer: hard to administer and secure at scale
Client/server: the server is a single point of failure and a cost

Source: Tutorial 1 (not on the slides)

CHAPTER 02

Physical layer

Chapter 2 · Parts 1 and 2: signals, transmission, guided and unguided media. 8 topics.

2.1

Digital transmission

Sending data as voltage pulses rather than a continuous wave.

Advantages

Low-cost technology
Longer distances using repeaters
Security and privacy (encryption)
Noise can be easily removed
Cheaper and easier multiplexing (many sources share one link)
Computers can process both analog data (voice, video) and digital data

Disadvantages

Most digital signals are aperiodic, so frequency doesn't describe them; bit rate is used instead

Source: Ch 2 Part 1 slides

2.2

Amplifier vs repeater (fixing attenuation)

Both fight attenuation; an amplifier for analog, a repeater for digital.

Advantages

Repeater: regenerates a clean copy of the digital signal, so it can travel farther
Amplifier: boosts a weakened analog signal

Disadvantages

Amplifier: boosts the noise along with the signal

Already discussedCh 1 Part 2: the repeater is listed as a physical-layer device.

Comes backCh 6: a hub also just repeats bits to every port.

Source: Ch 2 Part 1 slides (figure); tutorial

2.3

UTP (unshielded twisted pair)

Two insulated copper wires twisted together, no shield.

Advantages

Cheapest type of cable
Easiest to install
Twisting cancels interference from nearby pairs (more twists = better quality)
Lowest media and connector cost; lowest installation skill

Disadvantages

Suffers from external electromagnetic interference (EMI)
Relatively short distance: 1–100 m
Low immunity to EMI, radio-frequency interference and electrical hazards

Source: Ch 2 Part 2 slides

2.4

STP (shielded twisted pair)

Twisted pair with a metal braid or foil covering.

Advantages

The shield reduces interference

Disadvantages

More expensive
Harder to handle (thick, heavy)

Source: Ch 2 Part 2 slides

2.5

Fibre-optic cable

Glass or plastic core and cladding carrying light pulses.

Advantages

Higher bandwidth (hundreds of Gbps)
Smaller size and weight
Lower attenuation: greater repeater spacing (50 km or more without repeaters)
No crosstalk (no light leaking)
Not affected by noise; completely immune to EMI and electrical hazards
Highly secure (no light leaking)
Long distance: 1–100,000 m

Disadvantages

Needs special skills to install
More expensive interfaces than electrical ones; highest media and connector cost

Comes backCh 4: 10 Gigabit Ethernet runs on fibre only.

Source: Ch 2 Part 2 slides

2.6

Radio waves (< 300 MHz)

Omnidirectional wireless.

Advantages

Propagate in all directions, so one transmitter reaches everyone (multicast: AM/FM, TV)
Can travel long distances

Disadvantages

Available to everyone with a receiver (unguided media are not private)

Source: Ch 2 Part 2 slides

2.7

Microwaves (300 MHz – 300 GHz)

Unidirectional, line-of-sight wireless.

Advantages

Point-to-point links, e.g. between buildings' LANs
Used for unicast: cellular phones, satellite, wireless LANs

Disadvantages

Need line of sight (straight lines)
Very high frequencies cannot penetrate walls (a problem when receivers are inside buildings)

Source: Ch 2 Part 2 slides

2.8

Infrared (300 GHz – 400 THz)

Short-range, line-of-sight wireless.

Advantages

Very large bandwidth, not yet fully used
Stays in one room: good for remotes, wireless keyboards and mice (IrDA)

Disadvantages

Short distance only; cannot penetrate walls
Cannot be used under the sun (interference from the sun's infrared)
Line of sight needed

Source: Ch 2 Part 2 slides

CHAPTER 03

Data link layer

Chapter 3 · Part 1 (framing, error control) and Part 2 (media access control). 15 topics.

3.1

Simple parity check

One extra bit makes the number of 1s even (or odd).

Advantages

Detects every single-bit error
Very simple: one redundant bit

Disadvantages

Detects a burst only if an odd number of bits changed
Cannot tell which bit is wrong, so it cannot correct

Source: Ch 3 Part 1 slides

3.2

Two-dimensional parity

Parity for every row and every column of a block.

Advantages

Detects 1, 2 or 3 errors
Can correct a single-bit error (the failing row and column cross at it)

Disadvantages

Does not detect all 4-bit errors
More redundant bits than simple parity

Source: Ch 3 Part 1 slides

3.3

CRC (cyclic codes)

Modulo-2 division by a generator polynomial.

Advantages

Very good at detecting single-bit errors, double errors, odd numbers of errors and burst errors
Easy to implement in hardware and software; especially fast in hardware

Disadvantages

Bursts longer than the generator's degree can slip through (with small probability)
Detection only, no correction

Source: Ch 3 Part 1 slides

3.4

Checksum

One's-complement sum of the words, complemented.

Advantages

Simple to compute in software

Disadvantages

Weaker than CRC: some corruptions go undetected (e.g. all 1s, since one's complement has two zeros — Tutorial 4)

Comes backCh 7: the IPv4 header checksum. Ch 9: the UDP checksum.

Source: Ch 3 Part 1 slides; Tutorial 4

3.5

Error correction by retransmission

Detect the error, then ask the sender to resend.

Advantages

Simple: only detection bits are needed

Disadvantages

Inefficient on wireless links: the high bit error rate causes many retransmissions
Unacceptable delay for real-time multimedia: the application must wait for the resent packet

Comes backCh 9: TCP uses retransmission (reliable); UDP doesn't, which suits real-time apps.

Source: Ch 3 Part 1 slides

3.6

Forward error correction (FEC)

Correct a limited number of errors at the receiver (2-D parity, XOR, interleaving).

Advantages

No retransmission needed: suits wireless and real-time multimedia
Chunk interleaving spreads a burst so each packet loses only one chunk

Disadvantages

Can correct only a limited number of errors
Extra redundant data is sent every time

Source: Ch 3 Part 1 slides

3.7

Random access (CSMA family)

No station is superior; each decides when to send.

Advantages

Simple and easy to implement
Decentralized: no central device that can fail and bring down the system
Low delay in low traffic
A lone sender gets the full channel rate R

Disadvantages

In heavy traffic, packet delay has no limit
Unfair: a station may never get a chance to transmit
With M senders, throughput per node is less than R/M because of collisions

Source: Ch 3 Part 2 slides

3.8

CSMA (carrier sense)

Listen to the medium before transmitting.

Advantages

Reduces the possibility of collision

Disadvantages

Cannot eliminate collisions (propagation delay)
The longer the propagation delay, the worse the performance

Source: Ch 3 Part 2 slides

3.9

Non-persistent CSMA

Busy → wait a random time, then sense again.

Advantages

Random waiting reduces the probability of collisions

Disadvantages

Bandwidth is wasted if the backoff is long: the medium sits idle while stations wait

Source: Ch 3 Part 2 slides

3.10

1-persistent CSMA

Busy → keep listening; transmit with probability 1 as soon as it's idle.

Advantages

Avoids idle channel time

Disadvantages

Selfish: if two or more stations are waiting, a collision is guaranteed

Source: Ch 3 Part 2 slides

3.11

p-persistent CSMA

Idle → send with probability p, or wait one slot.

Advantages

Reduces collisions (like non-persistent)
Reduces idle time (like 1-persistent)

Disadvantages

Needs time slots equal to the maximum propagation delay

Source: Ch 3 Part 2 slides

3.12

CSMA/CD

Keep listening while sending; stop and jam on a collision.

Advantages

Stops a colliding transmission early, reducing channel wastage (plain CSMA keeps sending colliding frames)

Disadvantages

Only works where collisions can happen and be detected (shared medium, half duplex)

Comes backCh 4: Ethernet uses CSMA/CD, and full duplex makes it unnecessary.

Source: Ch 3 Part 2 slides

3.13

Controlled access (reservation, polling, token)

Stations take turns in an organized order.

Advantages

Fair: every station gets a chance; throughput R/M each
Eliminates collisions completely

Disadvantages

A node is limited to R/M; in centralized polling everything goes through the primary station

Already discussedCh 1: token passing circulates around a ring topology.

Source: Ch 3 Part 2 slides

3.14

FDMA / TDMA

Split the channel by frequency band or by time slot.

Advantages

No collisions: each station has its own band or slot

Disadvantages

A node is limited to R/M even when it is the only node with frames to send
TDMA: a node must wait for its slot even when the channel is free

Source: Ch 3 Part 2 slides

3.15

CDMA

One channel carries everyone at once; codes separate the stations.

Advantages

All stations transmit simultaneously over the full channel

Disadvantages

Each station's signal must be coded, and receivers must know the codes

Source: Ch 3 Part 2 slides

CHAPTER 04

Ethernet LANs

Chapter 4: Ethernet frames, addresses, standards and full duplex. 3 topics.

4.1

Traditional (half-duplex) Ethernet

Bus or star with a hub, using CSMA/CD.

Advantages

Simple shared medium

Disadvantages

Either transmit or receive, never both at once
Collisions waste bandwidth; CSMA/CD is required

Already discussedCh 3: CSMA/CD — its pros and cons are there. Ch 1: bus and star topologies.

Source: Ch 4 slides

4.2

Full-duplex (switched) Ethernet

Each station on its own point-to-point link to a switch.

Advantages

Throughput doubles (100 Mbps → 200 Mbps; 1 Gbps → 2 Gbps)
Each station is its own collision domain: no collisions, CSMA/CD no longer needed

Disadvantages

Needs full-duplex NICs, two wire pairs per link, and a switch (not a hub)
Devices must be connected point-to-point to the switch

Already discussedCh 1: full duplex as a transmission mode (telephone).

Comes backCh 6: switches and why they isolate collision domains.

Source: Ch 4 slides

4.3

Faster Ethernet standards

Fast (100 Mbps), Gigabit (1000 Mbps), 10 Gigabit (10,000 Mbps).

Advantages

Higher rates; Fast and Gigabit run on twisted pair or fibre

Disadvantages

10 Gigabit runs on fibre only (higher cost)

Already discussedCh 2: fibre's advantages and costs.

Source: Ch 4 slides

CHAPTER 06

Connecting devices & VLANs

Chapter 6: hubs, switches, routers and virtual LANs. 6 topics.

6.1

Hub (physical layer)

Copies received bits to every other port; no filtering.

Advantages

Extends the maximum distance between nodes (interconnects LAN segments)

Disadvantages

The whole hub is one collision domain (segments' collision domains merge), which reduces performance
No filtering: every frame goes everywhere
Bandwidth is shared: 100 Mbps / 4 hosts = 25 Mbps each

Already discussedCh 1: the hub is the centre of a star. Ch 2: a hub works like a multi-port repeater.

Source: Ch 6 slides

6.2

Link-layer switch

Forwards frames by destination MAC address, learning its table.

Advantages

Isolates collision domains (each port is separate): fewer collisions, higher total throughput
Increases performance and security: frames go only to the right port
Can work in full duplex (no CSMA/CD needed)
Increases geographical coverage; no limit on LAN size
Transparent / plug-and-play: no configuration needed

Disadvantages

Does not isolate broadcast domains: broadcasts reach the whole network
Floods frames for unknown destinations
Loops (redundant links) cause broadcast storms — needs the spanning tree protocol

Already discussedCh 4: full-duplex switched Ethernet doubles throughput.

Source: Ch 6 slides

6.3

Store-and-forward vs cut-through switching

Two ways a switch forwards frames.

Advantages

Store-and-forward: buffers the whole frame and performs error checking
Cut-through: forwards as soon as the destination address is read: no buffering delay

Disadvantages

Store-and-forward: buffering causes some delay
Cut-through: no error checking

Source: Ch 6 slides

6.4

Redundant links + spanning tree

Multiple paths between switches for reliability.

Advantages

Redundancy increases reliability; blocked ports take over if the network changes

Disadvantages

Without spanning tree, loops cause broadcast storms that consume all resources and stop the network

Source: Ch 6 slides

6.5

Router (network layer)

Forwards packets by destination IP address using a routing table.

Advantages

Isolates both collision domains and broadcast domains
Connects LANs and WANs with similar or different protocols
Chooses the best route using routing algorithms

Disadvantages

Needs configuration (not plug-and-play)
Drops packets whose destination isn't in its routing table
No cut-through forwarding

Comes backCh 7 and Ch 8: how routers forward and route packets.

Source: Ch 6 slides

6.6

VLAN (virtual LAN)

Broadcast domains (subnets) created in software, not by wiring.

Advantages

Reduces cost and installation time: move a station by software, not by rewiring
Increases security: a high-security group can be put in its own VLAN; broadcasts don't reach other VLANs
Creates virtual workgroups across physical locations (e.g. a cross-department project)

Disadvantages

Without VLANs, moving a user to another subnet means hard, costly rewiring

Source: Ch 6 slides

CHAPTER 07

Network layer: data transfer

Chapter 7: packet switching, IPv4 addressing, DHCP, NAT. 7 topics.

7.1

Virtual-circuit (connection-oriented) switching

A connection is set up first; all packets follow the same path.

Advantages

Packets of a message follow the same path (and arrive in order)
Resources such as buffer and bandwidth are set in every switch during setup

Disadvantages

Setup delay before data can be sent
Every packet must carry a virtual-circuit identifier (flow label)

Comes backCh 9: TCP is connection-oriented at the transport layer.

Source: Ch 7 slides

7.2

Datagram (connectionless) switching

Each packet is routed independently — the Internet's choice.

Advantages

No setup delay
Robust: if a router crashes, only the packets inside it are lost; others take other paths

Disadvantages

Packets are not guaranteed to arrive in the order they were sent
The full source and destination address must be attached to every packet

Comes backCh 9: UDP is connectionless at the transport layer.

Source: Ch 7 slides

7.3

Classless addressing (vs classful)

The mask, not the class, defines the network part.

Advantages

Solves the huge need for addresses and the IPv4 shortage
Addresses are distributed properly and saved; ISPs can subdivide their blocks (subnetting)

Disadvantages

Classful addressing wasted addresses: fixed block sizes (class A blocks of 16 million)

Source: Ch 7 slides

7.4

Subnetting

Dividing a network into smaller subnets with their own subnet addresses.

Advantages

Increases security (separates departments)
The site still looks like one network to the outside; outside routers use the main network address

Disadvantages

Two addresses per subnet (network and broadcast) can't be assigned to hosts

Source: Ch 7 slides

7.5

DHCP (vs manual configuration)

Hands out IP address, mask, gateway and DNS server dynamically.

Advantages

Works for diskless and first-time-booted computers
Handles computers that move between subnets
Leasing: addresses go back to the pool, so IP addresses aren't wasted

Disadvantages

Manual configuration (the alternative) must be changed by hand whenever a computer moves

Comes backCh 9: DHCP runs over UDP (port 67). Ch 10: it also hands out the DNS server address.

Source: Ch 7 slides

7.6

NAT

Maps private (local) addresses to the router's public address.

Advantages

Saves IP addresses: one public address for all devices
Change internal addresses without telling the outside world
Change ISP without changing internal addresses
Acts as a firewall: outsiders can't start a connection to your computer

Disadvantages

NAT will not allow more than one host to contact the same destination (per the slides)
Internet routers don't forward private destination addresses, so inside hosts aren't directly reachable

Source: Ch 7 slides

7.7

Directed broadcast

All host bits set to 1, sent to a specific remote network.

Advantages

Reaches every host on a remote network; routers forward it

Disadvantages

Usually blocked, because it can be used for denial-of-service attacks

Source: Ch 7 slides

CHAPTER 08

Network layer: routing

Chapter 8: routing tables, distance vector and link state, autonomous systems. 6 topics.

8.1

Static routing

Shortest paths computed offline and entered by the administrator.

Advantages

Simple; suits small networks

Disadvantages

Cannot update automatically when the network changes or fails
Entries must be typed in manually

Source: Ch 8 slides

8.2

Dynamic (adaptive) routing

Routers learn the network by talking to their neighbours.

Advantages

Each router fills its routing table automatically and adapts to changes

Disadvantages

More complexity is added to the router

Source: Ch 8 slides

8.3

Network-specific and default routing

Ways to keep routing tables small.

Advantages

Network-specific: one entry per network instead of per host
Default route 0.0.0.0/0 covers every destination not in the table

Disadvantages

Host-specific routing makes tables large

Already discussedCh 7: the default gateway on every host.

Source: Ch 8 slides

8.4

Distance vector (RIP, EIGRP)

Each router sends estimated costs to its neighbours.

Advantages

Each router only talks to its neighbours
Simple Bellman–Ford calculation

Disadvantages

Sends estimates, periodically (every 30 s) even when nothing changed
Reacts to link failure very slowly

Source: Ch 8 slides

8.5

Link state (OSPF)

Each router floods its exact link costs; every router builds the full map and runs Dijkstra.

Advantages

Sends exact link costs, and only when there is a change
Reacts to network failure quickly
Every router has the same complete map (LSDB)

Disadvantages

Information is flooded to all routers in the area
Every router must store the whole network and run Dijkstra

Source: Ch 8 slides

8.6

Hierarchical routing (autonomous systems)

Routing in steps: interior inside an AS, exterior (BGP) between ASs.

Advantages

Makes routing possible for the huge Internet
Each AS chooses its own interior protocol

Disadvantages

Border (gateway) routers must run both interior and exterior protocols

Source: Ch 8 slides

CHAPTER 09

Transport layer

Chapter 9: process-to-process delivery, TCP and UDP. 3 topics.

9.1

TCP

Connection-oriented, reliable byte stream.

Advantages

Reliable: segments arrive without error, loss or duplication, in order
Flow, error and congestion control
Suits applications that can't tolerate loss: WWW (HTTP), e-mail (SMTP), Telnet, FTP

Disadvantages

Connection setup (three-way handshake) and termination before and after data
Bigger header (20–60 bytes) and more overhead

Already discussedCh 7: virtual-circuit (connection-oriented) switching. Ch 3: retransmission-based error control.

Source: Ch 9 slides

9.2

UDP

Connectionless, unreliable datagrams.

Advantages

Simple, minimum overhead, no connection delay
Suits real-time apps (voice, video calls): a lost frame is just skipped
Suits one short request and response (DNS)

Disadvantages

Unreliable: no flow or error control; segments can be lost, duplicated, corrupted or out of order
On an error it simply discards the segment; the application must handle errors
Bad for long messages split into many datagrams (e.g. SMTP, file downloads)
A UDP server serves only one request at a time; the rest wait in a queue

Already discussedCh 7: datagram (connectionless) switching. Ch 3: why retransmission is bad for real-time traffic.

Source: Ch 9 slides

9.3

Connectionless vs connection-oriented service

Independence vs dependency between segments.

Advantages

Connectionless: no setup delay
Connection-oriented: data delivered in order, with flow, error and congestion control

Disadvantages

Connectionless: no ordering, no flow, error or congestion control
Connection-oriented: must establish and terminate the connection

Already discussedCh 7: the same split at the network layer (virtual circuit vs datagram).

Source: Ch 9 slides

CHAPTER 10

Application layer: DNS

Chapter 10: names, the DNS hierarchy and name resolution. 5 topics.

10.1

Host names (vs raw IP addresses)

People use names; DNS maps them to IP addresses.

Advantages

Names are easy to remember; IP addresses are not

Disadvantages

The network only understands numbers, so a lookup (DNS) is needed first

Source: Ch 10 slides

10.2

Centralized vs distributed DNS

Where the name database is stored.

Advantages

Distributed: many name servers in a hierarchy spread the load; no single host holds every mapping
Distributed: naming control is decentralized (each organization manages its own part)

Disadvantages

Centralized would be inefficient and unreliable: heavy traffic from all over the world, a failure makes the data unavailable, and it would be hard to maintain

Source: Ch 10 slides

10.3

DNS caching

Servers remember recent answers.

Advantages

Reduces search time for names outside the server's domain, increasing efficiency

Disadvantages

A cached answer can go stale if the record changes

Source: Ch 10 slides (advantage); the disadvantage is general knowledge

10.4

Primary vs secondary name servers

The primary loads the zone from disk; the secondary copies it (zone transfer).

Advantages

Secondaries share the load and keep answering if the primary fails

Disadvantages

Secondaries depend on zone transfers to stay up to date

Source: Ch 10 slides (roles); pros/cons follow from them

10.5

DNS over UDP

DNS uses port 53, normally over UDP.

Advantages

One short request and one short response fit in single datagrams: fast, no connection setup

Disadvantages

No built-in reliability (the resolver retries if no answer comes)

Already discussedCh 9: UDP's advantages and disadvantages.

Source: Ch 9 and Ch 10 slides