A Computer Network A Computer Network Computer Networks Computer - - PDF document

SLIDE 1

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 1

1

Computer Networks Computer Networks Part 1: Introduction Part 1: Introduction

Michael Welzl Michael Welzl http://www.welzl.at http://www.welzl.at DPS NSG Team DPS NSG Team http://dps.uibk.ac.at/nsg http://dps.uibk.ac.at/nsg Institute of Computer Science Institute of Computer Science University of Innsbruck University of Innsbruck

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 2

2

A Computer Network A Computer Network

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 3

3

Another Computer Network Another Computer Network

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 4

4

And this is...?! And this is...?!

Computer Network Distributed System

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 5

5

Data Communication vs. Data Communication vs. Computer Network vs. Distributed System Computer Network vs. Distributed System

• Data Communication

– how to transmit data between two connected computers

• Computer Network

– several connected nodes, some of which communicate :) – how to transmit data from computer 973476 to computer 8762876726 – some issues:

• medium sharing (5 computers, 1 cable?!)
• path finding (“routing“)
• ensuring reliability and security
• Distributed System

– Implies a certain level of abstraction – unaware of infrastructure (e.g., GRID) LAN MAN WAN

3 different things! don‘t care

• r even notice

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 6

6

LAN, MAN, WAN: quite different indeed! LAN, MAN, WAN: quite different indeed!

• Local Area Network (LAN) < 1.5 km

– now: (usually) customer premises – VPNs have no geographic boundaries!

• Metropolitan Area Network (MAN) - 2-50 km

– now: (usually) backbone of LANs – future: perhaps Ultrawideband / Wireless DSL connections?

• Wide Area Network (WAN) - country

– now: (usually) ISP Note the change: From definition to industry practice!

• Several other areas: car (e.g. Controller Area Network (CAN)),

home (e.g. CEBus), body (ubiquitous / wearable computing), interplanetary internet, ..

SLIDE 2

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 7

7

These things are real These things are real

http://www.ipnsig.org funded by DARPA NASA JPL involved http://www.media.mit.edu/wearables/ MIT Media Lab

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 8

8

Everything is totally heterogeneous Everything is totally heterogeneous

• Infrastructure

– copper, fibre, infrared, wireless

• Devices

– PC, notebook, PDA, cell phone, car display, sensor

• Services

– telephony, web surfing, email, chat, download, distributed computing – location-based / context-aware / ubiquitous services – note: should depend on devices

• surf with TV, cell phone: did not work well
• i-mode did (in Japan)
• video phone: old idea that never really made it
• Communication methods

– ad hoc networks, streaming, low-power comm., quantum cryptography, .. – note: should depend on devices and services!

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 9

9

Example: heterogeneous services Example: heterogeneous services

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 10

10

More realistic (?) ubicomp example More realistic (?) ubicomp example

• Infrastructure: wireless
• Device: PDA
• Communication Method: ad hoc networking
• Service:

– Two computer science students meet; they do not know each other. As they pass, Brook‘s PDA sets up an ad hoc network connection with Rich‘s PDA. The two devices exchange some information including a profile; Rich‘s PDA now knows that Brook is bad at maths and that she is generally interested in math related

• information. The device does not have the storage to permanently carry lecture

notes - besides, Rich usually writes them on plain paper - so it queries Rich‘s home database via the Internet about recent places Rich has visited (every once in a while, Rich‘s location is determined via his cell phone and saved in his home database). Rich‘s home server looks for Brook in its own database, where the names of girls like Tailor (who cheated on Rich) are stored. Brook is not in the database, so she might be all right; thus, it queries the university server for rooms and classes that match Rich‘s previous locations / times, and detects that Rich has just been to a Math class. It informs Rich‘s PDA, which provides this information to Brook‘s PDA. Her device now goes through the same process and notices that Rich has visited a class that Brook missed - so it hums softly. Brook smiles and says “hey, have we met before - and are you going to the party tonight?“

Hackers will like this ;-)

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 11

11

Example: Sensor Networks Example: Sensor Networks

• ACM SenSys‘03 - Call for Papers... “topics of interest include the following:“

– Network protocols for sensor networks – Operating system and middleware for sensor networks – Distributed database processing in sensor networks – Distributed algorithms for sensor networks – Novel sensor node hardware and software platforms – Sensor network planning and deployment – Energy management in sensor networks – Adaptive toplogy management – In-network processing and aggregation – Data storage in sensor networks – Distributed and collaborative signal processing – Distributed Actuation, Control, and Coordination – Localization in time and space – Distributed calibration in sensor networks – Simulation and optimization tools – Applications of distributed sensor networks – Security and Robustness in sensor networks – Sensor network testbed measurements and benchmarks

Classification attempt: Infrastructure Devices Services Communication methods

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 12

12

Dealing with heterogeneity Dealing with heterogeneity

• Networking

– used to be a new / young / fresh discipline – now approaching mid-age, but still somewhat immature

• Compare: “computer languages“ / “computer networks“

– well defined analytical background / plethora of modelling methods – well structured research area / somewhat disorganized research area – gone through several phases of maturity / still in infancy?

• Problem: constant flux

– now even with impact on “computer languages“ (.NET ⇒ C#) – changing infrastructure, devices, services, communication methods

• Long-lasting knowledge / invariants become increasingly important!

SLIDE 3

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 13

13

What invariants are there? What invariants are there?

• (Roughly) ordered according to longevity:

– General knowledge: rules / guidelines (KISS, ..) – Models (Graph Theory, Queuing Theory, FSM, ..) – Algorithms (Dijkstra shortest path first, ..) – How things are designed (how guidelines, models, algorithms are applied) – How things are deployed (standardization) – Certain technology that is difficult to change (TCP/IP, your TV, ..) – Human interaction with technology + economics (cell phones, ubicomp, ..) e.g. languages: (E)BNF - Logo

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 14

14

Some Invariants Some Invariants (“Theory“) (“Theory“)

ISO/OSI model, protocol, Message Sequence Charts, FSMs, connectionless vs. connection-oriented service, confirmed service functions, switching, routing, the Internet, DoD model, scalability, End2end Argument, Client-Server vs. Peer-to-Peer, IP hourglass

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 15

15

Most Important CN Invariant: Most Important CN Invariant: Abstraction Abstraction

• Programming Languages:

– Machine code – Assembler – Low level languages – High level languages (special purposes), OO Design, ..

• Simplification by abstraction
• Same for networks: Network Layer Models

– ISO/OSI model – DoD model

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 16

16

Logical communication flow Logical communication flow

draw a green rectangle

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 17

17

Physical communication flow Physical communication flow

draw a green rectangle Network (“Internet Cloud“) language (protocol) translation

010010101111010100101011101000101011000011110100010111001011

translation check for bit errors check for logical errors Choose a path

C H A O S

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 18

18

ISO/OSI Reference Model ISO/OSI Reference Model

• THE famous layer model
• 7 layers
• precise terminology, huge

amount of theoretical work

• layer provides service to upper

layers

• strict rules (layers must not be

skipped, but they are interchangeable)

Layer 2 Layer 1 Layer 0

request to secy.: request: transmit letter transport letter delivered, letter content please send 5 pairs of black suspenders service to higher layer according to “protocol” write order letter letter → opened, processed

UK US

delivered

SLIDE 4

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 19

19

OSI Architecture OSI Architecture

ISO Basic Reference Model for Open Systems Interconnection (OSI)

– OSI model is: concept, architecture, common terminology.

application layer presentation layer

session layer transport layer network layer data link layer physical layer 7 6 5 4 3 2 1 transport protocol transit network

Physical communication flow

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 20

20

Terminology: Protocol Terminology: Protocol, Service, , Service, Layer Layer

(Communication) Protocol:

• Communication based on messages
• needs rules + regulations, algorithmic description of collaborative

interaction between partners

• related to very different aspects (depending on layer) such as

– plug geometry, pins; voltage + signal form for “0” / “1”? – how to cope with transmission errors – how to find path to remote node – how to convert Sun-Unix-C “reals” to Pentium-Win98-VBX “reals”

(Communication) Service:

• entities communicate using protocol
• set of entities provides service (abstract machine) to “user“

Layer: grouping of (0,1,n) services with comparable functionality

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 21

21

Protocol Protocol, Service, , Service, Layer Layer (2) (2)

Terminology-at-a-glance:

• (N)-layer provides (N)-service to (N +1)-layer: “how” is hidden!
• “how”: (N)-entities communicate according to (N)-protocol; make

use of (N -1)-service provided by underlying layer;

• exception - lowest layer: actual message transport

node A node B

protocol Layer N+1 Layer N: services N1, N2 service interface entitiy vertical comm. horizontal comm. N1 N2 N1 N2 Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 22

22

Service Service Function Function, , Connection Connection

Many services grouped into 3 phases = 3 essential service functions:

1. Connection Establishment CON 2. Data Exchange DAT 3. Connection Release (Disconnect) DIS

“connection-oriented” CO services

• vs. “connectionless” CL services: Phase 2 (DAT) only

Fundamental distinction and “religious war” CO vs. CL: CO compares to “telephone” service, CL to “letter” postal service

• pro CO: much better “quality” of service (at cost of “memory”)
• pro CL: “stateless” scalable, charged by volume, not time; always-on!

net overloaded: CO – busy, no connection; CL – stalled, but no reject Beware: the tale of Internet overprovisioning!

CO important in the Internet, e.g. due to optical network technology

(N) Svc. Access Pt. (N)-ConnectionEndpt. (N)-connection (N)-service A user at node B B

Example: GPRS

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 23

23

CO vs. CL CO vs. CL

Basically, it all comes down to: ...how many calls can she handle? vs. ...what kind

• f guarantee

do you get? Note: CO can be built on top of CL service (i.e. she can call the post office to make sure that your packets have arrived - this happens on the Internet!)

(connection) switching packet switching, routing

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 24

24

Service Primitive, Service Primitive, Confirmed Confirmed Service Service

• Confirmed service functions: usually 4 service primitives:

1.Request (service user asks for service function) REQ 2.Indicate (service informs remote user) IND 3.Response (remote user replies back to service) RSP 4.Confirm (service reports back to initial user) CNF

• Unconfirmed service functions: just request indicate
• Usually: CON confirmed, DAT unconfirmed, DIS often confirmed

CON.REQ, CON.IND, CON.RSP, CON.CNF DAT.REQ, DAT.IND

• In „real“ protocols, names are usually different

(TCP‘s DAT.REQ: „send“, DAT.IND is call-return from „receive“)

SLIDE 5

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 25

25

Service Primitive, Finite State Service Primitive, Finite State Machine Machine

Description of Service Dynamics #1: Finite State Machine FSM

DisInd PAboInd1 DisReq PAboInd2 ConReq ConCnf ConRsp ConInd

.... .... ....

; PAboInd1,PAboInd2 ConReq; ; PAboInd1, PAboInd2 DisReq; DisInd PAboInd1

Further States (Phases Dat, Dis)

…

WAIT CONN- ECTED ConRsp; ConCnf ConReq; ConInd START

statics: FSM: note: not a classical FSM: a) non deterministic b) spontaneous transitions

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 26

26

Message Sequence Message Sequence Charts MSC Charts MSC

Description of Service Dynamics #2: Message Sequence Chart MSC

– state / time diagram – pro: simpler, intuitive; locations explicit – con: one MSC per alternative (# may be infinite!)

node j node k ConReq ConInd ConRsp ConCnf service Responder Initiator ConReq ConInd Responder Initiator PAboInd PAboInd

t t

service node j node k Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 27

27

OSI Model: OSI Model: Data Units Data Units 1 1

Application level “messages” - processed as data units. Common notions:

• packet: “unit of transportation” (may contain fragments)
• datagram: instead of packet if sent individually (connectionless)
• segment: instead of packet if part of a connection
• frame: with final envelope, ready to send (next to lowest layer)
• cell: small frame of fixed size

OSI terminology: „message“ is a PDU

• PDU: protocol data unit

– (N)-PDU: semantics understood by peer entities of (N)-service – (N)-PDU = (N)-PCI plus (N)-SDU; (N)-SDU = (N+1)-PCI plus (N+1)-SDU

• PCI: protocol control information: only used by peers
• SDU: service data unit - payload, optionally carried in PDU for user

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 28

28

H2 H3 H4 M1 T2 H2 H3 H4 M2 T2 H2 H3 H4 M T2 H2 H3 H4 M1 T2 H3 H4 M1 H3 H4 M2 H3 H4 M1 H4 M1 H4 M2 H4 M1 M M M M

6 5 4 3 2 1 7

10010111010011101010101101011 sent received → frame (packet with “envelope”, ready for transmission in (1)-Layer packet

OSI Model: OSI Model: Data Units Data Units 2 2

Header Hn (plus maybe Trailer Tn) in (N)-layer carries (N)-PCI Protocol Control Information PCI: checksums, msg no., …

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 29

29

OSI Layers OSI Layers

• Application oriented layers:
• 7. Application Layer: actual communicating apps
• 6. Presentation Layer: ensure conforming semantics
• 5. Session Layer: “structured dialogue“ - synchronization, half-

duplex vs. full-duplex, ..

• Transport oriented layers:
• 4. Transport Layer: end2end msg. stream, no knowledge of routing
• 3. Network Layer: routing, packets between adjacent systems

(LANs: 2b: Logical Link Control; L.2a: Medium Access Control)

• 2. Data Link Layer: error-recovery, frames (not packets) stream

between adjacent systems

• 1. Physical Layer: insecure bitstream between adjacent systems

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 30

30

The presentation problem The presentation problem

Q: Does perfect memory-to-memory copy solve “the communication problem”? A: Not always!

Problem: Different data format, storage conventions

struct { char code; int x; } test; test.x = 256; test.code=‘a’ a 00000001 00000011 test.code test.x host 1 format a 00000011 00000001 test.code test.x host 2 format

SLIDE 6

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 31

31

Solving the presentation problem Solving the presentation problem

• 1. Translate local-host format to host-independent format
• 2. Transmit data in host-independent format
• 3. Translate host-independent format to remote-host format

OSI host-independent format: “Abstract Syntax Notation One” (ASN.1)

defines “Basic Encoding Rules” (BER)

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 32

32

The largest The largest computer computer network is network is... ... (and (and the winner is the winner is...) ...) The The Internet Internet

• bvious source

for invariants!

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 33

33

How the How the Internet has Internet has changed changed

• Original goal: robust communication on a long-term basis

– military background (DoD) Well... (9/11)

• Original size: ARPANET...

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 34

34

How the How the Internet has Internet has changed changed

• Feb. 2001: 69 new hosts added each minute!

– [IEEE Spectrum]

• Commercial demand for new, accordingly priced services

(VoIP, streaming audio/video, videoconferencing, ..)

– Overprovisioning does not suffice - demands increase

• Goal is not just speed / reliability anymore;

Internet "best effort" service is not good enough

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 35

35

It still works quite well. It still works quite well.

Invariant ahead... ;-)

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 36

36

Primary reason for tremendous success Primary reason for tremendous success: : scalability scalability end2end argument:

leave application specific tasks up to applications the network should only do general (broadly useful) things ...important Internet design rule!

Today, scalability remains the no. 1 goal !

Example: security Security provided by the network may be too bad or unnecessary for some applications. The network cannot provide authentication: only the application has access to the necessary information

SLIDE 7

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 37

37

Some scalability hazards Some scalability hazards

• Per-flow state in the network

– possible solution: flow aggregation (reduce no. of flows)

• Traffic that scales like O(nx) instead of O(n)

(n ... number of participating nodes)

– Q: how to make traffic scale like O(n)? – A: Make it a function of n! (e.g., restrict signaling traffic to fixed percentage of data traffic)

• Bad traffic distribution

– DDoS attacks! Famous TCP SYN attacks also exploit per-flow state

• Central point of failure

– Classic client-server architecture does not scale well

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 38

38

Client Client-

• Server

Server

• Traditional model, easily comprehensible abstraction

– Clients request service (initiate connection) – Servers provide service (answer requests)

• Examples: Web Client/Server, Mail Client/Server, FTP Client/Server

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 39

39

Peer Peer-

• to

to-

• Peer

Peer

Peer-to-peer:

• Brand new paradigm
• Very successful since 00‘s

– 230,309,616 Kazaa downloads (May 26, 2003) - world record!

• First tool: Napster (file sharing)

– route around censorship

• Other services

– Streaming media – distributed computing

Other cool P2P technologies: ;-)

• Telephone
• Usenet
• DNS
• IP Routing

Actually, P2P = original Internet model (symmetric network - all hosts are (ftp, telnet, ..) clients and servers

?

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 40

40

Common Internet Common Internet myths myths + + misconceptions misconceptions

• The network automatically finds the best (uncongested) path for my packets.
• IPv6 is brand new, provides QoS guarantees and will be widely deployed soon.
• Overprovisioning can solve all QoS problems.
• World Wide Web = Internet.
• The Internet exists since ...
• TCP/IP is (are) the Internet protocol(s).
• Somewhat less common:

Internet2 will soon replace the Internet as we know it.

• The Internet changes with incredible speed.

Rather, “p2p works“ is an invariant!

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 41

41

DoD reference model DoD reference model

• DoD-model older (approx. 1969) than OSI (approx. 1977)
• Only 4 layers:

– Application Layer – Transport Layer – Internet Layer (= layer 3, Network layer) – Network Access Layer (= everything underneath the Internet)

• Layers 5 and 6 missing
• Less restrictive - layers may be skipped

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 42

42

OSI, DoD and reality OSI, DoD and reality

• OSI:

Many service functions carried out in several layers / services Overhead, even reversal!

• “Why should I implement layers 5, 6 in my app / OS ?“
• Commercial failure - but still useful to explain networks
• DOD:

actually obsolete; Internet defined by TCP/IP stack (defined by RFCs)

SLIDE 8

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 43

43

TCP/IP Protocol Stack TCP/IP Protocol Stack

• IP: addressing, routing, fragmentation/reassembly, TTL
• UDP: ports, checksum
• TCP: UDP + connection-oriented service (retransmission/ACK),

combined flow control / congestion control

Physical IP TCP UDP HTTP,FTP,.. Application Transport Network

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 44

44

A shaky invariant: the Internet Hourglass A shaky invariant: the Internet Hourglass

Everything Over IP IP Over Everything

No No assu assumption mptions ⇒ no gu no guarante arantees! es!

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 45

45

W Who defines the Internet? ho defines the Internet?

• Preliminary research in IRTF ( http://www.irtf.org )

consisting of Research Groups

• Standards (RFCs) defined by IETF ( http://www.ietf.org ) -

mostly Working Groups

• Decisions by IESG (as of Feb. 2001, 14 elected members)
• IAB stimulates IETF / IESG actions

– Members elected by “Internet Society“ (ISOC)

• RFCs have different status: standard, proposed standard, draft standard,

experimental, informational, ..

• Internet-draft: preliminary - may turn into RFC

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 46

46

W Who ho really really defines the Internet? defines the Internet?

• RFCs define the Internet. Only standard RFCs mandatory
• Extremer views, but maybe closer to reality:
• Jon Postel defined the Internet.
• Cisco defines the (core) Internet.
• Microsoft defines the (end2end) Internet.
• Some standards never made it.

Some other things did (MBone, NAT, ..)

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 47

47

Standards Standards that never that never really really made it made it (and (and probably never probably never will) will)

• TTL as an actual time value
• The (originally planned usage of the) IP TOS field
• Several IP options:

Strict / Loose Source Routing Record Route Timestamp

• ICMP Source Quench

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 48

48

Who runs the Who runs the Internet? Internet?

• (See http://www.caida.org

for more pictures)

• Hierarchical structure
• Distributed; ISP admins run it
• No Internet "police"
• Each ISP interested in its own

services, but end2end paths may include the backbone

• CPU cycles scarce in core routers

simulated transit-stub network (ns/nam dump)

SLIDE 9

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 49

49

Internet: not the only network! Internet: not the only network!

The Internet dominates global communications - true / false ?

• Some US statistics:

– Business + consumer ISP market: revenue $13B each (2000) – Revenues in other industries: TV broadcast $29.8B (1997), cable distribution $35.0B (1997), radio broadcast $10.6B (1997), phone $268.5B (1999) – Internet reaches 59% of US households - vs. Telephone 94%, TV 98%

Telecom World: International Telecommunications Union ITU

– represents “common carriers” (AT&T…), PTTs (Telecoms …) – ITU-R: “radio communication” (former CCIR) – ITU-T: “telecommunications” (former CCITT) – ITU-D: development – standards usually as letter-stop-digits. letter=“series” (V.24 part of V-series: data transmission over analog lines) – Europe: CEPT = ITU-Europe; ETSI: standard body by ITU + industry!

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 50

50

Non Non-

• Internet standardization bodies 2

Internet standardization bodies 2

International Standards: Intl. Standards Organization ISO

– represents some 90 member countries (US: ANSI, GY: DIN …) – TCs, SCs, WGs (technical and sub-committees, working groups) – branch-spanning “expert groups” like MPEG – stepwise standardization CD → DIS → IS – ITU + ISO often cooperate (X.400 = ISO MOTIS, …)

US: ANSI represents ISO interfers or cooperates with

• NIST (Dep. of Commerce)
• DoD-Standards,
• IEEE standards

(e.g., IEEE 802 LAN standards have become ISO 8802)

• many others

effort started February 1980

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 51

51

Things to come Things to come

• HOW + WHY (long-lasting knowledge) more important than WHAT
• 1. Acquire overview of current technology (WHAT):

– Roughly follow layers top-down – Consider differences in heterogeneous networks

• 2. Fully understand concepts (HOW+WHY):

– Follow layers bottom-up – In-depth description of concepts / ideas – e.g.: “How does... the Internet find a path for my packets?“ short long

Uni Innsbruck Uni Innsbruck Informatik Informatik -

• 52

52

References References

• Some slides from:

– Max Mühlhäuser, Raj Jain, Andrew Tanenbaum, James Kurose, Keith Ross

• Some content from:

– Andrew Tanenbaum, "Computer Networks", Prentice Hall, 2002, 4. Edition, ISBN: 0130661023 – Bruce S. Davie, Larry L. Peterson, David Clark, "Computer Networks: A Systems Approach", Morgan Kaufmann; 3rd edition, 2003, ISBN: 1558605142 – James F. Kurose, Keith W. Ross, "Computer Networking. A Top- Down Approach Featuring the Internet", Addison Wesley Publishing Company, 2000, ISBN: 0201477114 – P. Molinero-Fernandez, N. McKeown, H. Zhang, “Is IP going to take over the workd (of communications)?“, ACM Computer Communications Review 33(1), Jan 2003.

• Some pictures from:

– Atlas of Cyberspace: http://www.cybergeography.org/atlas/atlas.html – Quirit: http://www.quirit.com – http://www.ietf.org/proceedings/01aug/slides/plenary-1/index.html

• Original ubicomp student exchange idea by Alois Ferscha