Advanced Internet Technologies - Computer and Communication

Advanced Internet Technologies, SS 2004 1.1

Advanced Internet TechnologiesChapter 1

Introduction

Dr.-Ing. Falko Dressler

Chair for Computer Networks & InternetWilhelm-Schickard-Institute for Computer Science

University of Tübingen

http://net.informatik.uni-tuebingen.de/[email protected]


Falko Dressler

1990-1998: Study of Computer Science, University of Erlangen-Nuremberg, Germany

1998:M.Sc. in Computer Science (Dipl. Inf.), Department of Computer Science, University of Erlangen-Nuremberg, GermanyThesis Title: Monitoring of ATM Networks (Netzmonitoring auf ATM-Ebene)

1999-2003: Ph.D. studies of Computer Science, Department of Computer Science, University of Erlangen-Nuremberg, Germany

2003:Ph.D. in Computer Science (Dr.-Ing.), Department of Computer Science, University of Erlangen-Nuremberg, GermanyThesis Title: Monitoring of Multicast Networks for Time-Synchronous Communication

Since 2003: Senior Researcher (wiss. Assistent), Computer Networks and Internet, Wilhelm-Schickard-Institute for Computer Science, University of Tuebingen, Germany


Course organization

LectureThursday, 15.00-17.00, A301

Questions and AnswersDr.-Ing. Falko Dressler

After the [email protected]

News and updateshttp://net.informatik.uni-tuebingen.de/

SlidesAll slides are available online (and without the slides it is difficult to follow the lectures...)! The slides may be updated during the course.


Chapter 1Introduction

General Course BibliographyHistory of the InternetOSI Reference ModelTCP/IP SuiteIP, IPv6, TCP, UDPRouting Protocols


General Course Bibliography

[Bla00] U. Black: IP Routing Protocols. Prentice Hall, 2000.[Com00] D. E. Comer: Internetworking with TCP/IP, Vol.1: Principles, Protocols, and

Architectures. Prentice Hall, 2000.[Fer98] P. Ferguson, G. Huston: Quality of Service. John Wiley & Sons, 1998.[Hän91] R. Händel, M. Huber: Integrated Broadband Networks. Addison Wesley, 1991.[Hui99] C. Huitema: Routing in the Internet. 2nd Edition, Prentice Hall, 1999.[Pry96] M. DePrycker: Asynchronous Transfer Mode. Ellis Howood/Prentice Hall, 1996.[Sta01] W. Stallings. High-Speed Networks and Internets: Performance and Quality of

Service. Prentice Hall, 2nd ed, 2001.[Ste94] W. R. Stevens: TCP/IP Illustrated, Vol.1: The Protocols. Addison-Wesley,

1994.[Ste95] W. R. Stevens: TCP/IP Illustrated, Vol.2: The Implementation. Addison-Wesley,

1995.[Ste96] W. R. Stevens: TCP/IP Illustrated, Vol.3: TCP for Transactions, HTTP, NNTP,

and the UNIX Domain Protocols. Addison-Wesley, 1996.


History

Situation in the 70sProprietary Protocols (e.g. DECNET, SNA)

GoalsOpen Communication PlatformReliable Data Communication

Development1966: Development of a packet network by DARPA (Defense Advanced Research Projects Agency)Concurrent developments at ”Xerox Palo Alto Research Center“ (PARC)1969: First presentation of the ARPANET1980: Development of TCP/IP1983: TCP/IP became standard for all computers connected to the ARPANET1984: Full TCP/IP support for 4.2BSD and UNIX System VToday: Billions of devices interconnected through the Internet


History II


Standardization

IAB: Internet Activities BoardIETF: Internet Engineering Task ForceIRTF: Internet Research Task Force

RFC: Internet Request for Comments

IAB

IESG IRSG

IETF IRTF

WG RG ......


OSI Reference Model

ApplicationProvides access to the OSI environment for users and also provides distributed information services

PresentationProvides independence to the application processes from differences in data representations (syntax)

SessionProvides the control structure for communication between applications; establishes, manages, and terminates connections (sessions) between cooperating applications

TransportProvides reliable, transparent transfer of data between end points; provides end-to-end error recovery and flow control

NetworkProvides upper layers with independence from the data transmissions and switching technologies used to connect systems; responsible for establishing, maintaining, and terminating connections

Data LinkProvides for the reliable transfer of information across the physical link; sends blocks (frames) with the necessary synchronization, error control, and flow control

PhysicalConcerned with transmission of unstructured bit stream over physical medium; deals with the mechanical, electrical, functional, and procedural characteristics to access the physical medium


TCP/IP Reference Model

Application

Presentation

Session

Transport

Network

Data Link

Physical

OS

I

Link

Application

Transport

Network

TCP/

IP

Link LayerAccess to the physical medium, packaging of the data

Network LayerAddressing, routing, fragmentation

Transport LayerEnd-to-end data transport, reliable transmission, flow control

Application LayerApplication support

TCP/IP Example


Host A Host B

Application

Transport

Link

Network

ftp

TCP

Ethernet

IP

Router

Link

NetworkIP Protocol

Link Protocol Link Protocol

IP Protocol

Transport protocol

Application protocol

Network 1 Network 2 (Ethernet)


TCP/IP Protocol Suite

TCP/IP consists of many sub protocols. Therefore, it is called a protocol suite.

Application

Transport

Network

Link

ApplicationProcess

ApplicationProcess

TCP UDP

ICMP(v4v6) IP(v4v6) ARP RARP

Hardware Interface


Internet Datagram

Attention: IP works connection less, i.e. every packet travels the network independently of its predecessors and its successors. Itmay happen that packets are fragmented. In this case, every fragment is processed like an individual IP packet. Therefore, the sequence of the packets might be changed within the network.

Host B

Host A

Packet 1

Fragment 2

Fragment 1


IPv4 Header0 VERS

IdentificationTime To Live Protocol

Service TypeHLEN Total LengthFragment OffsetFGS

Header Checksum

Destination IP AddressIP Options

DataPadding

Source IP Address

4

8121620

24

Legend:• VERS: version (4)• HLEN: header length in 32 bit words• Service Type• Total Length: packet length in bytes (max. 65535)• Identification: unambiguous identification of the packet• FGS: fragmentation flags• Fragment Offset: offset of the current fragment in 8 byte words• Time To Live (TTL): lifetime of the packet• Protocol: transport layer protocol• Header Checksum• Source IP Address• Destination IP Address• IP Options• Data: payload


IPv6 Header

IPv6 Header Header Extension 1

Header Extension n Payload…

optional

VersionPayload Length

Source Address

Destination Address

Next Header Hop LimitTraffic Class Flow Label

GoalsMoving options to extension headersSimplification of the headerElimination of some attributes (header length, header checksum)


ICMP (Internet Control Message Protocol)

The Internet Control Message Protocol, ICMP, allows gateways to send error or control messages to other gateways or hosts.ICMP packets are standard IP packets with protocol set to 1. The information consists of a type field and a code field.

Code Value

Description

0 Network unreachable1 Host unreachable2 Protocol unreachable3 Port Unreachable4 Fragmentation needed, DF set5 Source route failed6 Destination network unknown7 Destination host unknown8 Source host isolated9 Communication with destination network prohibited10 Communication with destination host prohibited11 Network unreachable for type of service12 Host unreachable for type of service

Type Field

Description

0 Echo Reply3 Destination Unreachable4 Source Quench5 Redirect8 Echo Request11 Time Exceeded12 Parameter Problem13 Timestamp Request14 Timestamp Reply17 Address Mask Request18 Address Mask Reply


TCP (Transmission Control Protocol)

RFC 793Connection-oriented transport protocolReliable data transferFlow controlCommonly used in wide area networks

TCP offers a data stream interface which can be used by applicationsSocket concept: a socket allows the unambiguous identification of a service on a host

Formal: a connection (association) is described by a 5-tupel:Protocol (TCP)Local IP addressLocal process (port)Remote IP addressRemote process (port)

Example: {tcp; 134.2.11.157; 1022; 134.2.11.152; 22}


TCP Header

Source Port: source processDestination Port: destination processSequence Number / Acknowledge NumberHLEN: header length in 32 bit wordsFlags

URG: urgent pointer significantACK: acknowledgment field significantPSH: push functionRST: reset the connectionSYN: synchronize sequence numbersFIN: no more data from sender

Window: current window sizeChecksum: error detection codeUrgent Pointer: informs about urgent data upcoming the stream

Sequence Number

HLEN Reserved

Source Port Destination Port

Acknowledge Number

Urgent PointerOptions

DataPadding

WindowFlagsChecksum


TCP – History of a Connection

Host 2

Receive SYN + ACK segment

Send ACK y+1

Connection Release

Receive SYN segmentSend SYN seq=y, ACK x+1

Send FIN seq=x‘

Receive ACK segment

Receive FIN + ACK segmentSend ACK y‘+1

Receive FIN segmentSend ACK x‘+1

Send FIN ACK x‘+1

Receive ACK segment

Inform Application

Receive ACK x+2

Send Packet x+1Data Transfer

Send SYN seq=xConnection Establishment

Host 1

Receive ACK segment

Receive Packet x+1Send ACK x+2


TCP – Finite State Machine

Finite State Machines are employed for the specification and verification of protocols.They consist of states (spheres) and state transitions (arrows).

closed

listen

SYNRECVD

SYNSEND

established

FINWAIT-1

FINWAIT-2

TIMEDWAIT

closing

CLOSEWAIT

LASTACK

fin/ack

fin/ack

fin/ack ackack ack

close/finclose/fin

ack

reset

passive open closeactive open/syn

send/syn

syn/syn+ack close/timeout/reset

begin

syn/syn+ack

fin/ack

close/fin

ack

timeout nach 2 „segment lifetimes“


UDP (User Datagram Protocol)

RFC 768Connectionless transport protocolUnreliable service (delivery and duplicate protection are not guaranteed)No flow controlMore efficient and robust

Typical application scenariosInward data collection (active or passive sampling of data sources such as sensors)Outward data dissemination (distribution of real-time clock values)Real-time applications (e.g. voice and telemetry)

UDP headerSource port, destination portSegment length, checksum


Routing

What is Routing?Forwarding of packets between different local networksSpecial hosts, named routers, determine the next hop for each arriving packet based on lookups in their routing tables

What is Routing not?The calculation of (optimum) paths through the networkThe routing protocol is used to determine paths through the network based on information known via static route entries or dynamic route updatesThe routing protocol is employed to create and maintain the routing table

(3) (5)

Host A Host BRouter

(1), (2)(4)


Routing Protocols

Static RoutesRouting tables are created and maintained by network administratorsVery simpleEasily to understandWorks for simple topologiesNo redundancy

Dynamic Routing ProtocolsRouter exchange information about the network topologyExtra resources are required

Network bandwidth for route updatesCPU power for route calculations

Works even for complex topologies (depending on the algorithm)Allows redundancy and load balancing


Autonomous Systems

Set of Routers under centralized / common administration

NW

NWNW

NWNWNW

NW

AS 1

Area

Area

Area

Area

AS 3

AS 4

AS 2


Internal / External Gateway Protocols

IGP – within an autonomous systemEGP – between autonomous systems

EGP

IGP

AS 1AS 2

IGP

IGP

IGP

IGPIGP

IGP

IGP

IGP

IGP


Single-Homed (Stub) AS / Network

Stub network: only a single connection to the global networkRoutes from the backbone to the stub commonly via static routesRoutes from the stub to the backbone commonly via a default gateway

NW 1

NW 2

NW 3

NW 4ISPAS 1

default


Multi-Homed Non-Transit AS / Network

NW 1

NW 2

NW 3

NW 4

NW 5

NW 6

ISP B

ISP A

AS 1

NW 1, NW 2

NW 3, NW 4

NW 1, NW 2

NW 5, NW 6


Multi-Homed Transit AS / Network

NW 1

NW 2

NW 3

NW 4

NW 5

NW 6

ISP B

ISP A

AS 1

NW 1, NW 2,NW 5, NW 6

NW 3, NW 4

NW 1, NW 2,NW 3, NW 4

NW 5, NW 6


Distance Vector Protocols

PropertiesAlso called minimum hop protocolsVery simpleEasy to maintain

Working principlesEvery router maintains a table with best paths to every known destinationThis table is periodically distributed to each neighbor

ProblemsSlow convergence in case of route changesOnly a single metric for optimum path selection (distance)


Distance Vector Protocols II

G F

C

E

D

BA

I am next to G

I am 3 to GI am 2 to G

I am 2 to G

I am 1 to G

I am 5 to G

I am 4 to GI am 3 to G


Link State Protocols

PropertiesEach link has an associated metric (cost factor)The metric can be available bandwidth, monetary costs, link speed etc.

Working principleEvery router has a database with complete information about every router, its interfaces and the associated metricsPeriodically, the database information is distributed to all other routersThe routing table is calculated using the database and a shortest path algorithm

ProblemsHigh resource requirements (memory, CPU)


Link State Protocols II

Conclusion:I am 9 to G through B

G F

C

E

D

BA

F: G=2, E=1

C: E=2, D=2, A=5E: F=1, C=2F: G=2, E=1

E: F=1, C=2F: G=2, E=1

2

2

2

1

5

1

B: D=1, A=1D: C=2, B=1C: E=2, D=2, A=5E: F=1, C=2F: G=2, E=1

Conclusion:I am 10 to G through C

C: E=2, D=2, A=5E: F=1, C=2F: G=2, E=1

D: C=2, B=1C: E=2, D=2, A=5E: F=1, C=2F: G=2, E=1

1


SPF (Shortest Path First)

Also called “optimum path”Terminology

Node: bridge, gateway, routerPath: connection between two nodes

Least-cost weighting based onDelayThroughputConnectivity

Well-known example: Dijkstra algorithm1. Let D(v)=sum of link weights on a given path2. Let c(i,j)=the cost between node i and j3. Set n={1}4. For each node (v) not in N, set D(v)=c(1,v)5. For each step, find a node w not in N for which D(w) is a minimum; add w to set N6. Update D(v) for all nodes still not in N by: D(v)=min[D(v), D(w)+c(w,v)]7. Repeat steps (4) through (6) until all nodes are in set N


Dijkstra Example

2

8

4

14

81

51

4

7

2

2

4 3

A G

I

E J

HF

C

D

B

A C-4 B-8

B A-8 D-2 F-4

C A-4 D-1 I-7 E-4

D B-2 C-1 E-2

E C-4 D-2 F-4 G-2

F B-4 E-4 H-3

G E-2 I-1 H-1

H F-3 G-1 J-8

I C-7 G-1 J-5

J H-8 I-5


Dijkstra Example II(4,A)

2

8

4

14

81

51

4

∞7

4 3

2

2A G

I

E J

HF

C

D

B

workingnode

2

8

4

14

81

51

4

4 3

7

2

2A G

I

E J

HF

C

D

B

(8,A)

∞∞

∞ ∞

∞

∞∞

workingnode

(4,A) (11,C)

(8,C)

(8,A)

∞ ∞

(5,C)

∞∞


Dijkstra Example III(4,A)

2

8

414

81

51

4

7

4 3

2

2A G

I

E J

HF

C

D

B

(5,C)

∞

(11,C)

workingnode

(7,D) ∞

∞∞


Dijkstra Example IV

The result of the SPF algorithm is a spanning tree topology with the following capabilities:

There is a path to every other nodeThere are no loops in the tree

Remember: there is a different spanning tree for every node the algorithm is started on

2

14

1

51

42

2A G

I

E J

HF

C

D

B

(7)(1)

(2) (3) (5)

(9)

(6)

(4) (8)


RIP – Routing Information Protocol

RFC 1058Developed by Ford und Fulkerson; also known as Ford-Fulkerson algorithmVery simple distance vector protocolHop count is the only metric

RIPv1 does not support variable length netmasks (Classless Interdomain Routing, CIDR)

Working principleMaximum hop count: 16 (infinity)Distribution of network addresses and distances (hop count)Periodically broadcast (every 30 sec)180 sec without a new update: route is marked unusable240 sec without a new update: route is removed


RIP – Packet Format

RIPv1Bit 0-7 Bit 8-15 Bit 16-31

Command Version All zeroes

Address family All zeroes

IP address

All zeroes

All zeroes

Metric

repetition of previous 20 bytes

repetition of previous 20 bytes

Metric

Next hop IP address

Subnet mask

IP address

Route tagAddress family

Routing domainVersionCommand

Bit 16-31Bit 8-15Bit 0-7

RIPv2


RIP – Example / Convergence

(2,5)

(1,5)

Network 6 Network 5

Network 4Network 3

Network 2

(3,5)

(4,5)

R2

R5

R3

R4

R1

RIP routing table at R3Destination Next hop Metric Direct or Remote Local or RIP Interface

Network2 0 1 D L 2

Network3 R2 2 R R 2

Network4 0 1 D L 1

Network5 R5 2 R R 1

Network6 R2 3 R R 2


RIP – Counting to Infinity: the Problem

R1 distributed a route to Host1Some failure happens at H1R1 distributes a new route to H1 using a metric of 16 (infinity)R2 distributes an update just before it gets the new information from R1R1 receives this update and

Discovers that there is a path to H1 through R2 with metric 2R1 creates a new entry in its routing table for H1 with metric 3R1 sends an update to its neighbors

Therefore, R1 and R2 slowly increase the metric in their routing tables for H1this holds if R1 and R2 reach metric 16the convergence time is very highunnecessary network load is generated

R1

R2 R3

Network 1

Failure

Network 2


RIP – Counting to Infinity: the Solution

Split HorizonKey idea: it is useless to announce routes in the same direction from where the announcements have been received

Split Horizon with Poison ReverseKey idea: send information towards the direction from where an announcement has been received as a route update with metric 16 (infinity)

Hold-downKey idea: for a period of time do not accept new routes to a destination for which an update has been sent be oneself


OSPF – Open Shortest Path First

Developed by J. MoyOSPFv1: RFC 1131, 1245, 1246OSPFv2: RFC 1247, 1253Link state protocol

AdvantagesSupport for flexible metrics for routing decisions (e.g. distance, throughput, current load, $$)Large range of values for the metric (1 to 65535)Scalability (2-level hierarchy)Support for multiple paths (equal path load sharing, unequal path load sharing)Support for asymmetric metrics (based on directed graphs)Support for secured messages (MD5 checksum)


OSPF – Basic Operations

Router A Router B

Hello

Hello

Merely Adjacent Merely AdjacentDatabase Descriptions

Database Descriptions

Fully Adjacent Fully AdjacentLink State Updates

Link State Updates


OSPF – Flooding of Routing Information

(1) Initial advertisement

R3 decides to send LSAs to its neighbors

R2

R5

R6

R3

R1

R4


OSPF – Flooding of Routing Information II

(2) Following advertisements

R2

R5

R6

R3

R1

R4


OSPF – Flooding of Routing Information III

(3) Acknowledgements

Note 1: ACKs are not required if a LSA was already sentNote 2: Multiple LSAs are not transmitted, multiple ACKs are transmitted

R2

R5

R6

R3

R1

R4Note 1

Note 1

Note 2

Note 2


OSPF – LSA Header Format

LS Age (16)

Options (8)

LS Type (8)

Link State ID (32)

Advertising Router (32)

LS Sequence Number (32)

LS Checksum (16)

Length (16)


OSPF – LSA Format

Header (160)

Router Type (8)

Reserved (8)

Number of Links (16)

Link ID (32)

Link Data (32)

Link Type (32)

TOS Metrics (8)

Metric (8)

Repetitionfor eachconnectedlink


OSPF – LSA Example

172.16.1.6

172.16.1.1

172.16.1.3 172.16.1.5

172.16.1.4172.16.1.2

6

5

1

2

Header

Router Type = 0

Number of Links = 3Link ID = 172.16.1.3

Link Data = ifIndex: 2

Link Type = Pt-to-Pt: 1

TOS Metrics = 0

Metric = 6

Link ID = 172.16.1.2

Link Data = ifIndex: 1

Link Type: Pt-to-Pt: 1

TOS Metrics = 0

Metric = 5Link ID = 172.16.1.1

Link Data = (mask)

Link Type = Stub: 3

TOS Metrics = 0

Metric = 0


OSPF – Areas

Problem in very large networksThe number of LSAs increases dramatically, therefore, the network load is increased as well

SolutionEstablishment of multiple areasEstablishment of designated routers (DRs)Prevention of the requirement that each router has to have a complete knowledge about the complete network

AdvantagesReduction of the size of the local database at each routerReduction of the number of LSAs in the network


OSPF – Designated Router

OSPF defines a DR for every networkElection based on the highest IP addressElection may be manipulated by using administrative prioritiesTypically, IP multicast is employed for the election (group 224.0.0.5)

In point-to-point topologies, every router has exactly one neighborIn broadcast networks with n deployed OSPF routers, every router has n x (n -1)/2 neighbors

The DR is employed to reduce the number of neighbor ship relationsTo increase the reliability, a backup DB (BDR) is elected


OSPF – Designated Router II

R5 R6

R1 R3

R2 R4

R5 R6

R1 R3

R2 R4

n x (n-1)/2 neighbor ship relations

DR

n-1 neighbor ship relations

Documents

Advanced Internet Technologies - Computer and Communication