06

Purposes l What is routing control?

Introduction to routing ∼Basics of designing the networks ∼

l Why is routing control necessary? l How can routing control be effectively applied for network design?

21, December, 2000

Internet Initiative Japan, Inc. Jiro Yamaguchi ([email protected])

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

1

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

2

Notation of a network

Contents l Roles of a data link layer and a network layer l Differences among hubs, switches, and routers l Static routing and dynamic routing l The operating principles of dynamic routing l backup and balancing using dynamic routing

H

l Network design l Address allocation policies

R

H

l A hub or a switch is described, using a single line. l A host is described as H, and a router is described as R. l In the explanation, Layer 3 switches are not discriminated from routers.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

3

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

4

OSI reference model and TCP /IP

Data link frame and routing l The roles of a data link layer and a network layer are explained.

OSI reference model ７ ６ ５ ４ ３ ２ １

l Why are the addresses for both MAC address (Ethernet address) and IP address required? l Why is routing necessary? l Why can communications be achieved without routing?

Application layer Presentation layer Session layer Transport layer Network layer data link layer Physical layer

TCP /IP HTTP, SMTP, etc.   TCP and UDP IP Ethernet, FDDI, ATM, etc.

ＯＳＩlayer Layer 2 ：Data link layer Layer 3 ：Network layer 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

5

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

6

1

IP datagram which flows on Ethernet

Connected network l A and B are directly connected to the same network.

C

– Possesses the correspondence table of MAC address and IP address by ARP (address resolution protocol), etc. ↓

D To C

Ethernet

l This is called “connected.” ↓

To A

A

B Recipient

l No needs to set up routing. Communications can be achieved when a hub and the like are connected.

Sender

A’s MAC address B’s MAC address

Header

FCS

Data

Sender

Recipient

B’s IP address

A’s IP address

Data link frame

Frame Check Sequence

connected

A

B

IP datagram

Header

Network X

Data

R

Not connected

Network Y C 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

7

2000/12/21

D Copyright © 2000 Internet Initiative Japan Inc.

8

How does a connected network look like from network layer -１

How does a connected network look like from network layer -２

l Address of A

l Address of A

– 192.168.1.1/24

– 192.168.1.1/24

l Connected address space from the viewpoint of A

l Connected address space from the viewpoint of A

– 192.168.1.0 to 192.168.1.255

– 192.168.1.0 to 192.168.1.255

l Allocate the address to B, selecting from 192.168.1.2 to 192.168.1.254

l Allocate the address to B, excluding the ones from 192.168.1.2 to 192.168.1.254

– Allocate 192.168.1.2 to B – Communications can be achieved between A and B A

Communicable

– No communications can be achieved between A and B Uncommunicable A B

B 192.168.1.0/24

192.168.1.1/24 2000/12/21

192.168.1.2/24

192.168.1.1/24

Copyright © 2000 Internet Initiative Japan Inc.

9

Unconnected network -１

Copyright © 2000 Internet Initiative Japan Inc.

l Set up routing

– A: Route network Y to R – C: Route network X to R

l By doing so, mutual communications between A and C can be achieved.

l Without setting up routing, no communications can be achieved between A and C.

– R is connected with both A and C, therefore, communications can be achieved when addresses are set up.

Routing table of A

Routing table of C

Destination Next Hop Reachability X Connected Reachable Y No Unr

Destination Next Hop Reachability X No Unr Y Connected Reachable

Routing table of A Destination Next Hop Reachability X Connected Reachable Y R Reachable

[訳注：1]

B

A Communicable

R

2000/12/21

Network X

Network Y

network X

D Copyright © 2000 Internet Initiative Japan Inc.

B

R

Network Y C

Routing table of C Destination Next Hop Reachability X R Reachable Y Connected Reachable

network Y

Network X Uncommunicable

10

Unconnected network -２

l As A and C are separately connected to different networks, they are unconnected.

A

2000/12/21

172.16.1.1/16

C 11

2000/12/21

D Copyright © 2000 Internet Initiative Japan Inc.

12

2

The status of the data link frame

Summary of the network terminology l Destination, or Recipient Address

MAC address IP address

A XA XR

E-Dest

E-S r c

IP -S r c

IA IR０

XR

XA

IA

IC

Data

FCS

IR1 IC

YC

YR

IA

IC

Data

FCS

– Means a destination. In a network, it is literally handled as the destination address, or the recipient address. “Destination” is frequently used as what it is, without being translated into Japanese. In the case of routing, “Destination” refers to the network information which includes mask information as well as address.

IP -Dest

R YR YC

C

l NEXT HOP, and NEXT HOP Address – The next address to which packets are forwarded. When a router or a host is neither “Destination”, nor “Connected”, the next address to send a packet (NEXT HOP) is referred in order to send the IP packet. The router or the host which receives the IP packet forwards it to its next address (NEXT HOP). This is repeated to reach the “Destination.”

IP datagram Ethernet data link frame

l The recipient and sender of the IP datagram never change on the way.

l Routing, Routing information

l The data link frame changes whenever it passes a router.

l Routing Table

l The ”data link frame recipient ” does not always mean the "IP datagram recipient. "

l Route

2000/12/21

– Means a route. Destination and NEXT HOP are paired.

Copyright © 2000 Internet Initiative Japan Inc.

– List of routings that a router and/or a host possesses – The Status where a router normally sends out an IP packet, following the routing table. “This router routes correctly. ” " 13

The status of the data link frame XA XR

C

14

l Destination, or Recipient Address E-Dest

E-S r c

IP -S r c

IA IR０

XR

XA

IA

IC

data

FCS

IR1 IC

YC

YR

IA

IC

data

FCS

– Means a destination. In a network, it is literally handled as the destination address, or the recipient address. “Destination” is frequently used as what it is, without being translated into Japanese. In the case of routing, “Destination” refers to the network information which includes mask information as well as address.

IP -Dest

R YR YC

Copyright © 2000 Internet Initiative Japan Inc.

Summary of the network terminology

MAC address IP address

A

2000/12/21

l NEXT HOP, and NEXT HOP Address – The next address to which packets are forwarded. When a router or a host is neither “Destination”, nor “Connected”, the next address to send a packet (NEXT HOP) is referred in order to send the IP packet. The router or the host which receives the IP packet forwards it to its next address (NEXT HOP). This is repeated to reach the “Destination.”

IP datagram Ethernet data link frame

l Routing, Routing information

l The recipient and sender of the IP datagram never change on the way. l The data link frame changes whenever it passes a router. l The ”data link frame recipient ” does not always mean the "IP datagram recipient. "

– Means a route. Destination and NEXT HOP are paired.

l Routing Table – List of routings that a router and/or a host possesses

l Route – The Status where a router normally sends out an IP packet, following the routing table. “This router routes correctly. ” "

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

15

Summary of data link frame and routing l When a data link layer as well as a network layer are “Connected”, communications can be achieved without setting up routing.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

Functional differences between switches and routers l Functional differences between hubs and switches

l When a network and a host is “Unconnected”, a router and routing always need to be set up.

l Effective use of switches

l The recipient and sender of the IP datagram never change on the way.

l Automatic set-up of a network

l The data link frame changes whenever it passes a router.

l Fault tolerance of switches

Copyright © 2000 Internet Initiative Japan Inc.

l Set up to use routers l Differences between switches and routers

l The ”data link frame recipient ” does not always mean the "IP datagram recipient. 2000/12/21

16

l Fault tolerance of routers l Broadcast flood 17

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

18

3

Differences between hubs and switches -1

Effective use of switches

Constructed using a hub

Server

A

100BaseTX

B

Switch Hub 8 ports x 10BaseT

To A

A C

l Mainly effective for the traffic between a server and a host. l A ⇔ Server

l Therefore, it propagates the communications between different ports to other irrelevant ports, interfering other communications. Copyright © 2000 Internet Initiative Japan Inc.

H ⇔ Server

19

Setting to use a router Server Network A

Network A,B

Network A,C

Network B A

B

– Automatically sets up routing. – Mainly used between routers. – RIP, RIP 2, OSPF, and the like – Automatically selects a backup route when a failure occurs.

l Set up the routing of the network on the other end of the communications. – it can be automatized by protocols such as DHCP and dynamic routing protocol.

Copyright © 2000 Internet Initiative Japan Inc.

21

Differences between switches and routers C

Network A

20

l Dynamic routing protocol

l A network is divided into subnets.

Server

Copyright © 2000 Internet Initiative Japan Inc.

l DHCP (Dynamic Host Configuration Protocol) – Automatically allocates addresses. – RFC2131 – Mainly used for a client. – Automatically renumbers, therefore, it possesses portability.

Network C

R Network A,C

2000/12/21

Each of them can fully use 10BaseT

Automatic setset-up of a network

C Network B,C

2000/12/21

H

D

l All the ports are continuously connected to the hub.

2000/12/21

B

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

22

Functional differences between switches and routers l Functional differences between hubs and switches

Network C

– Switches don’t propagate the communications of different ports to other ports.

R

l Differences between switches and routers – Routers don’t propagate the communication between different networks to other networks. – Different from switches, routing needs to be set up. – Needs to divide the network into subnets.

Network B A

B

l The router doesn’t propagate the communications between certain networks to other irrelevant networks.

l To use switches effectively – Introduce switches to the port on which the traffic concentrates. What are the problems?

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

23

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

24

4

Fault tolerance of switches -1 Server

Fault tolerance of switches -2

C

Network A

Server Network C

C

Network A

Network C All hosts can ’t get

Switch

Switch

access to the server

Network B A

B

Network B

←The identical IP

A

B

address with the server

←The identical IP address with the server

If the identical IP address with

l When switches are used, the wrong setting at one client causes a network-wide problem.

the server is allocated to B by mistake … 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

25

Fault tolerance of routers -1 Server

2000/12/21

Server Network C

C

Network A

Network C

Only B can’t communicate

R

26

Fault tolerance of routers -2

C

Network A

Copyright © 2000 Internet Initiative Japan Inc.

The network B only can’t communicate with the server

R

with the server Network B A

B

Network B

←The identical IP

A

B

address with the server l When routers are used, the wrong setting at one client doesn’t cause a network-wide problem.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

l When routers are used, even in the worst scenario, the impact made by the wrong setting at one client remains within the segment. 27

Broadcast Flood -1 Server

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

C

Broadcast packet has the impact

packet to other networks

Broadcast

B

A

l When the number of hosts grows, it creates the traffic that the broadcast packet can not ignore.

Copyright © 2000 Internet Initiative Japan Inc.

B

l No broadcast flood arises.

l Windows OS tends to create such broadcast packets in high volume.

2000/12/21

The router doesn’ t pass the broadcast

R

on all ports connected with switches

Broadcast

A

28

Broadcast Flood -2 Server

C

Switch

←The identical IP address with the router

l Supports large-scale networks.

29

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

30

5

Connection to the Internet

Switches VS Router – No needs to consider routing. – When compared to hubs, a more efficient network can be constructed.

Routing by ISP

Communicable Routing by ISP

default

l Advantages of routers

l Default means – Terminology in computer and internet industries. – The route which is selected when no specific routes are selected. – Different from the financial term of “default.”

R

– Backup can be constructed using dynamic routing protocol. – No broadcast flood arises. – Scalable even when the network size grows. – Can minimize the damage inflicted by a fault. – Relatively easy switching operation when a fault occurs.

default A

l Conclusions – Divide the network into subnets by routers, and introduce switches to the ports on which traffic concentrates on.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

31

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

l Characteristics of static routing

l Characteristics of static routing (static) and dynamic routing (dynamic) l Operating principles of dynamic routing

– Manually set up a route in a dormant manner. – Stable. – Has no impact made by traffic and transmission failures. – Creates no traffic derived from routing protocols.

l Types and characteristics of dynamic routing l What is RIP?

l Characteristics of dynamic routing – – – –

l VLSM l What is OSPF? l Trouble shooting 33

Reasons to choose dynamic routing -1

Automatically sets up a route. Can respond to the changes of the network. Can automatically select the optimized route. Can automatically select the backup route.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

Internet Headquarters R

– To prevent rewriting all when a network is added

R R R

l Needs to connect the networks whose organizers are different

Branch A

– Connection with multiple administrated networks

R

l Facilitates the setting of routers

35

2000/12/21

Branch B

Branch C R

R R

– Difficult to manually control a large-scale network Copyright © 2000 Internet Initiative Japan Inc.

34

Intricately expanding network

l Must respond to the changes of the network

2000/12/21

32

Static routing and dynamic routing

l Principles of dynamic routing are explained

Copyright © 2000 Internet Initiative Japan Inc.

l By directing the default route to the router which is connected with the Internet, communications with servers on the Internet can be achieved. l Routing is indispensable for the connection with the Internet

Explanation about routing

2000/12/21

default WWW Server, etc. Server

Internet

l Advantages of switches

R R

R

Copyright © 2000 Internet Initiative Japan Inc.

36

6

Reasons to choose dynamic routing -2

Backup between Tokyo and Osaka

Internet

lCan automatically select the optimized route.

Internet Tokyo

– Complicated network topology out of control.

R

lCan automatically select the backup route.

Osaka

Leased line

R

R

R

– When the network which needs to defend to the last exists. – Consider the structure which defies failures.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

37

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

Dynamic routing: routing information after propagation

Dynamic routing: propagation of routing information Internet

Internet

R

R

default

PC

2000/12/21

PC

Copyright © 2000 Internet Initiative Japan Inc.

…

default route

PC

39

Types of dynamic routing protocols

2000/12/21

PC

…

Copyright © 2000 Internet Initiative Japan Inc.

40

RIP l Routing Information Protocol Version 1

lRIP

l RFC 1058

–RFC 1058

l Propagates only addresses

lRIP ２

– Can be used for VLSM

–RFC 2453

l Vector -distance routing

lOSPF –RFC 2328

l Broadcast only

lBGP 4

l Included in UNIX as standard (route D)

–RFC 1771

2000/12/21

38

Copyright © 2000 Internet Initiative Japan Inc.

41

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

42

7

RIP2

OSPF--1 OSPF

l Routing Information Protocol Version 2

l Open shortest path first

l RFC 2453

l RFC 2328

l Can propagate netｍask

l Protocol 89

– can be used for VLSM

– Neither TCP (protocol 6) nor UDP (protocol 17)

l Vector-distance routing l Compatible with RIP, and can be used concurrently

l Can propagate net mask – Can be used for VLSM

l Can use multicast – To reduce the burdens of a host

l Recently supported by some routers［訳注：2］ 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

43

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

OSPF--2 OSPF

BGP4--1 BGP4

l Uses multicast (224.0.0.5 /224.0.0.6)

l Border Gateway Protocol Version 4

l Implements load -balancing

l RFC 1771

l Not included in UNIX as standard

l TCP 179

– Needs to install gated, etc.

44

l EBGP as EGP, and IBGP as IGP l Selects a route in accordance with the length of the AS path

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

45

BGP4--2 BGP4

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

46

What is dynamic routing? l Understand RIP – When RIP is understood, it is easy to understand the concepts of OSPF and BGP 4.

l Propagates using only the optimized route when more than two routes exist

l In the fields, RIP is still used in some cases

l Doesn’t implement load -balancing

– Because the routers for which OSPF can not be applied still exist – Because RIP is sufficient enough when only default is sent.

l Update protocol l Can aggregate, and supports Classless Inter-Domain Routing (CIDR)

l What is OSPF? – Will be explained based on RIP.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

47

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

48

8

The distance increments by one whenever data pass a router

RIP operating principles -1 Vector-distance routing (vector-distance /Bellman-Ford)

192.168.1.0

Vector=destination (network ) Distance=HOP count (the number of routers that the data pass)

RIP

RIP

R

R

Dest=192.168.1.0 Dist= 0

R

Dest=192.168.1.0 Dist= 1

Dest=192.168.1.0 Dist= 2

Dest=Destination Dist= Distance

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

49

When the destination is the same, the route with shortest distance is Shorter one is selected selected. Dist= 0

Dist= 1

Dist= 2

RIP

192.168.1.0

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

Broadcast once every 30 seconds 192.168.4.0 Once every 30 seconds

R

RIP

50

1

192.168.2.0

Not selected

2

3

4

5

Dist= 3

RIP

RIP

RIP Dist= 1

192.168.4.0

192.168.4.0 192.168.4.0 192.168.4.0

192.168.2.1

192.168.2.1 192.168.2.1 192.168.2.1

Dist= 2

When the destinations and the distances are the same, the priority is given to the route which is achieved first. 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

51

The route is deleted when its routing information is not provided in 3 minutes

R

3

4

5

192.168.4.0

192.168.4.0 192.168.4.0 192.168.4.0

192.168.2.1

192.168.2.1 192.168.2.1 192.168.2.1

Copyright © 2000 Internet Initiative Japan Inc.

52

RIP operating principles -2 l When a fault occurs in the network, a route is switched in 3 minutes. When multiple routers exist, it takes for 3 minutes X the number of routers network.

Fault occurs!! → 180 seconds later

2

2000/12/21

l RIP doesn’t propagate net mask. l Is considered to be classful mask. – Can be used when the address is

Deleted

Deleted

Deleted

Deleted

n

192.168.1.0/24

n

172.16.0.0/16

n

10.0.0.0/8

The routing information obtained by RIP is 180 seconds 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

53

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

54

9

RIP propagation

Routing information after RIP propagation

Internet R1

192.168.4.0

Internet 192.168.2.0 192.168.3.0 192.168.4.0

default

default

R4 192.168.4.0/24

R2

192.168.3.0

192.168.2.0/24

192.168.1.0/24 192.168.1.0 192.168.4.0

192.168.4.0

R4

default

192.168.2.0 192.168.3.0

default

default

192.168.1.0/24 192.168.1.0 192.168.4.0

192.168.4.0

192.168.2.0 192.168.3.0

R1

R2

default

192.168.2.0 192.168.3.0

192.168.4.0/24

R3

192.168.2.0/24

R3

192.168.3.0 192.168.3.0/24

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

55

RIP operating principles -3

192.168.3.0/24

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

56

Summary of RIP -1 l Vector-distance routing (vector-distance /bellman ford)

l Can not be used when the address is

– Vector = destination (network ) – Distance = hop count (number of routers that the data pass)

– 192.168.1.0/26

l The distance increments by one whenever data pass a router.

– 172.16.0.0/24

l The address of 0.0.0.0 serves as default.

l When the destination is the same, the route with shortest distance is selected. l When the destinations and the distances are the same, the priority is given to the route which is achieved first.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

57

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

58

Network configuration with subnet mask

Summary of RIP -2

Internet

l Broadcasts every 30 seconds. 192.168.2.192

l Deletes the route whose routing information has not been provided for 3 minutes.

default 3 193 192.168.2.192/26

– When multiple routers exist, it takes for 3 minutes X the number of routers. Copyright © 2000 Internet Initiative Japan Inc.

1

default

192.168.2.0/26

192.168.2.192 2

R4

l When a fault occurs in the network, route is switched in 3 minutes.

2000/12/21

192.168.2.64 192.168.2.128

R1

192.168.2.0 192.168.2.192

R2 192.168.2.64 192.168.2.128 192.168.2.128

default

65 66

192.168.2.64/26

R3 192.168.2.128/26

59

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

60

10

To use subnet mask by To use subnet mask by RIP -1

To use subnet mask by RIP -2

l Apply the net mask which is set up in the interface.

lWhen the net mask set up by the interface can not be used, RIP can not control routing.

l When the 192.168.2.1/26 router address is masked. Recipient table obtained by RIP 192.168.2.64 192.168.2.65 192.168.2.128 192.168.2.192 192.168.3.0 192.168.3.64

2000/12/21

Routing table 192.168.2.64/26 192.168.2.65/32 192.168.2.128/26 192.168.2.192/ 26 192.168.3.0/24 192.168.3.64/32

Copyright © 2000 Internet Initiative Japan Inc.

61

Network configuration with VLSM

R

PC 192.168.5.65 Not propagated

2000/12/21

192.168.5.128/ 25

– To support VLSM, use RIP 2 or OSPF

63

Routing control by RIP at a router

l

Request

2000/12/21

Can be used only to advertise default information Use when the default is not advertised

Copyright © 2000 Internet Initiative Japan Inc.

Copyright © 2000 Internet Initiative Japan Inc.

64

l When the identical broadcast address is not used – When the Broadcast addresses are different – When 192.168.1.0/24 is used n 192.168.1.255 network+all-1 n 192.168.1.0 network+all-0 n 255.255.255.255 all-1 n 0.0.0.0 all-0

Can be operated by RIP alone

-

2000/12/21

Trouble shooting: RIP doesn't propagate -1

Advertisement

-

l When 192.168.5.1 receives 192.168.5.128 – Confused with 192.168.5.128/26 – From 192.168.5.192 to 192.168.5.255, no routing is made. l VLSM can’t be supported by RIP alone

PC 192.168.5.193

Copyright © 2000 Internet Initiative Japan Inc.

62

l Examples of networks – 192.168.5.0/26 – 192.168.5.64/26 – 192.168.5.128/25

192.168.5.128 192.168.5.0/2 6 192.168.5.128 192.168.6.0 R 192.168.5.128 192.168.6.0/2 192.168.5.64 4

R

Copyright © 2000 Internet Initiative Japan Inc.

VLSM (Variable Length Subnet Mask)

R

192.168.5.64/2 6

2000/12/21

l Some old routers and workstations use either 0 or 1 for all.

65

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

66

11

Backup using RIP - Route propagation (under normal conditions)

Trouble shooting - RIP doesn't propagate -2 l The broadcast address is filtered – Does the interface output filter off at 255.255.255.255 or 0.0.0.0?

Network A A-Dist: ０ B-Dist: ３

Network B

R

l The protocol or the port is filtered – Is UDP 520 filtered?

Network B

Network A

Network A

l Broadcast could not propagated via unnumbered interface – Set up to advertise via unicast. – Is it O.K. to advertise using unicast?

Network A-Dist: １Network A A-Dist: ２ B-Dist: ２ B-Dist: 1Network B A-Dist: ３ Main circuit R B-Dist: ０ R

R A-Dist: １ B-Dist: ３

Network B Network A Sub circuit

Network A

B

R

Network A Network B Network A

R

R

A-Dist: ２ A-Dist: ３ Network BB-Dist: ２Network B B-Dist: １

Due to its distances are greater than the other, these are not selected l The configuration uses RIP, and mainly aims at backup. l Under normal conditions, only main circuit is used.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

67

Main circuit

R

Network B

R Network A

R

Network A

R

R R

R

Network A

R

Main circuit

R Network B

Network A Sub circuit

Network B

Network B

Network B

Network A

68

Network B

Network A

R

Copyright © 2000 Internet Initiative Japan Inc.

Backup using RIP - Traffic flows (under normal conditions)

Backup using RIP - Routing table (under normal conditions) Network A

2000/12/21

R

R

Network A

Sub circuit

R

R

Network B

Network B

l Due to the propagation of RIP routing information, the routing information is set up at respective routers.

l Under normal conditions, only main circuit is used.

l Due to the difference in distance, the main circuit route is selected. 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

69

Backup using RIP: Route propagation (when a failure occurs) A-Dist: １ B-Dist: ４ Network A

Failure

A-Dist: ５ B-Dist: 1

R

R

A-Dist: ０ B-Dist: ４ Network B Network A

Network A

R A-Dist: １ B-Dist: ３

Network A Sub circuit

Network B

Network B

Network A

R Network B

R

Network A

Network A

A-Dist: ２Network BA-Dist: ３ Network B B-Dist: ２ B-Dist: １

Copyright © 2000 Internet Initiative Japan Inc.

Network B R

R

R

Network A Network B Network A

R

70

Failure Network A

R

Network A Sub circuit Network B

l As a failure occurs on the main circuit, the propagation of the routing information changes.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

Backup using RIP: Routing table (when a failure occurs)

NetworkA-Dist: B ４ B-Dist: ０

Network A

R

2000/12/21

R

Network A Network B Network A

R

R Network B

l Due to the changes of the routing information propagation, the routing information set at respective routers changes.

71

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

72

12

What is OSPF? -１

Backup using RIP - Traffic flows (when a failure occurs) Failure

Network B

l Policies for this explanation

R

R

R

– General applications will be explained to those who don’t know OSPF. – Some may differ from the strict definitions about OSPF defined by RFC, however, that is to give better and easy-to-understand pictures to you. Your understanding is greatly appreciated. – For a large -scale network, the association with BGP is indispensable, but, it is not explained this time.

R Sub circuit

R

Network A

R

R

l As a failure occurs on the main circuit, the traffic flow changes. l The sub circuit is used as backup to maintain communications. 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

73

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

74

What is OSPF? -２

What is OSPF cost?

l Link State type routing protocol

l OSPF uses “Cost”, which is equivalent with “Distance” of RIP.

– Creates the database of the network topology in the format called LSA (Link State Advertisement) to select the optimized route. n

– The OSPF cost value varies from 0 to 65535. – Cost can be set up for respective interfaces, as desired. – The smaller cost means smaller distance. – Some routers automatically add costs, depending on the line speed, but, it may not be able to support the speedup of the network. Therefore, it is safe to explicitly set up the important interfaces including backbone.

Different from RIP and BGP, simple route exchange is not implemented, therefore, routing filter is difficult to implement.

– When the topology changes, immediately, the change is reflected. – Can detect a broken router. n n

Using HELLO packets, a broken router is detected to switch to the backup route. Switching is remarkably faster than RIP (for several seconds to approximately 1 minutes).

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

75

Simple way to calculate OSPF cost -1 Cost to Network A 0

R1

Circuit

Cost to Network B 86

R3

Cost: 66

Cost: 10

H2

Cost is set up at each I/F Cost: 10

76

Circuit

Cost: 66

Cost to Network B 0

Network B Network B

R2

R3

Cost: 10

H2

Cost: 10

l Route to H2 from R3

– R1 is directly connected to Network A, and the cost of H1 which is also connected to Network A is considered to be zero.

– R3 is directly connected to Network B, and the cost of H2 which is also connected to Network B is considered to be zero.

l Route to H1 from R2

l Route to H2 from R2

– From R2, the cost will be: [the cost of Network A which is set up at R1I/F] + [the cost of the I/F which is connected to R1]

– From R2, the cost will be: [the cost of Network B which is set up at R3 I/F] + [the cost of the I/F which is connected to R3]

l Route to H1 from R3

l Route to H2 from R1

– From R3, the cost will be: [the cost of Network A from R2] + [the cost of the I/F which is connected to R2] Copyright © 2000 Internet Initiative Japan Inc.

R1

Cost is set up at each I/F

l Route to H1 from R1

2000/12/21

H1

Cost to Network B 20

Network B

Network A

Network B

R2

Copyright © 2000 Internet Initiative Japan Inc.

Simple way to calculate OSPF cost -2

Cost to Network A 86

Network A

Network A

Network A

H1

Cost to Network A 76

2000/12/21

– From R1, the cost will be: [the cost of Network B from R2] + [the cost of the I/F which is connected to R2] 77

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

78

13

In order to implement backup and balancing

Simple way to calculate OSPF cost -３ Network A: 0 Network B: 86

Cost is set up at each I/F

Network B

Circuit

R1

Network A: 86 Network B: 0 Network A,B

Network A,B

Network A

H1

Network A: 76 Network B: 20

R2

R3

H2

l OSPF can afford backup and balancing when it has multiple routes. l When routes have different costs

Cost: 10

Cost: 66

Cost: 66

Cost: 10

Cost: 10

– The route with smaller costs can be used as a main route, and the the one with greater costs can be used as backup.

Cost: 10

l By assigning the same cost to the same I/F, the costs for outgoing and return can be identical.

l When routes have the same costs

l Different costs can be separately assigned for outgoing and return, but this will make the control complicated. Therefore, it should not be implemented without some particular reasons.

– By balancing, the traffic can be dispersed. – Even if one of the route for which balancing is implemented, remaining routes can be serve as backups.

l The figure here may give you the impression that routes are exchanged, but, practically, the route is determined by exchanging topology database. 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

79

Backup using OSPF -Route propagation (under normal conditions) Main circuit Cost: 10 A-Cost(R1): 20 B-Cost(R3): 86

Cost: 10

A-Cost(R1): 0 B-Cost(R2): 96 B-Cost(R5): 163

R1

R3

R2

Cost: 66

Set the cost at a greater number

R5 Network A

A-Cost(R2): 86 B-Cost(R4): 20

Cost: 133 A-Cost(R3): 96 A-Cost(R5):153 B-Cost(R4): 20

B-Cost(R4): 0 A-Cost(R3): 96 A-Cost(R6):163 Network B

l When a failure occurs, the sub circuit is used as backup.

Cost: 10

R2

Router name of the propagation source (NEXT HOP)

81

R3

R1

Network A

R5

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

Cost: 10

Cost: 10

Traffic to Network A Traffic to Network B

Copyright © 2000 Internet Initiative Japan Inc.

82

Main circuit

B-Cost(R4): 0 Network A-Cost(R6):163 B

A-Cost(R1): 0 B-Cost(R5):163

R2

A-Cost(R6): 163 B-Cost(R4): 20 Cost: 66

R3

R1 R6

Network A A-Cost(R1): 20 B-Cost(R6):153

Cost: 133 A-Cost(R5):153 Cost: 10 B-Cost(R4): 20

Sub circuit Cost value Router name of the propagation source (NEXT HOP)

83

l The backup is completed using the sub circuit.

2000/12/21

B-Cost(R4): 0 Network A-Cost(R6):163 B

R4 R5

Cost: 10

Sub circuit l When the line is cut off, the connection between R2 and R3 is deleted.

2000/12/21

R6

Cost: 133 A-Cost(R3): 96 A-Cost(R5):153 B-Cost(R4): 20

Cost: 10 A-Cost(R1): 20 B-Cost(R6): 163

Cost: 10

Cost: 10

B-Cost(R4): 0 A-Cost(R3): 96 A-Cost(R6): 163 Network B

R4

A-Cost(R1): 20 B-Cost(R2): 96 B-Cost(R6):153

R6

Cost: 133 A-Cost(R5):153 B-Cost(R4): 20

R3

l As OSPF HELLO packets flow in the sub circuit as well, it is impossible to make its traffic zero.

R4

A-Cost(R1): 20 B-Cost(R6):153

Cost: 66

Backup using OSPF -Traffic flows (when a failure occurs)

A-Cost(R6): 163 B-Cost(R4): 20 Cost: 66

A-Cost(R2): 86 B-Cost(R4): 20

Sub circuit

Cost value

Copyright © 2000 Internet Initiative Japan Inc.

Cost: 10 A-Cost(R1): 20 B-Cost(R6): 163

R2

R1

backup using OSPF -Route propagation (when a failure occurs) Main circuit

Network A

A-Cost(R1): 0 B-Cost(R2): 96 B-Cost(R5): 163

Route which is not selected

Sub circuit

A-Cost(R1): 0 B-Cost(R5):163

Cost: 10 A-Cost(R1): 20 B-Cost(R3): 86

Cost: 10

R5 Cost: 10

80

Backup using OSPF -Traffic flows (under normal conditions) Main circuit

R4

l Using OSPF, only the main circuit is used under normal conditions.

Cost: 10

Copyright © 2000 Internet Initiative Japan Inc.

R6

A-Cost(R1): 20 B-Cost(R2): 96 B-Cost(R6):153

2000/12/21

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

Cost: 10

Traffic to Network A Traffic to Network B

84

14

Characteristics of OSPF backup routing l Different from RIP, quick backup can be accomplished.

A-Cost(R1): 0 B-Cost(R5):163

85

Characteristics of OSPF backup routing l Different from RIP, quick backup can be accomplished.

R1

R2

Adjust to the same cost

R5 Network A

R3

A-Cost(R1):20 B-Cost(R6):86

Cost: 66

Cost: 10 A-Cost(R1):20 B-Cost(R3):86 B-Cost(R5):96

R2

Cost: 66

R3

Adjust to the same cost

Cost: 66

A-Cost(R1):20 B-Cost(R2):96 B-Cost(R6):86

l From R4 to Network A, the costs of both R3 and R6 should be the same. 2000/12/21

B-Cost(R4): 0 Network A-Cost(R3):96 B A-Cost(R6):96

R4

R6 A-Cost(R3):96 A-Cost(R5):86 B-Cost(R4):20

l From R1 to Network B, the costs of both R2 and R5 should be the same.

Cost: 10

Cost: 10

Route which is not selected Cost value Router name of the propagation source (NEXT HOP)

Copyright © 2000 Internet Initiative Japan Inc.

Cost: 10 A-Cost(R1):20 B-Cost(R3):86 B-Cost(R5):96

Cost: 10 B-Cost(R4): 0 Network A-Cost(R6):96 B

A-Cost(R1): 0 B-Cost(R2):96 B-Cost(R5):96

R4

R1

R2

88

Cost: 10

Network A

A-Cost(R2):86 A-Cost(R6):96 B-Cost(R4):20 Cost: 66

R3

Adjust to the same cost

R5 Cost: 10

A-Cost(R2):86 A-Cost(R6):96 B-Cost(R4):20

l Set up the two lines at the same costs.

R6 A-Cost(R5):86 B-Cost(R4):20

86

Backup and balancing using OSPF -Traffic flows (under normal conditions)

A-Cost(R6):96 B-Cost(R4):20 Cost: 66

Traffic to Network B

normal conditions)

R5

Backup and balancing using OSPFOSPF-Route propagation (when a failure occurs)

A-Cost(R1): 0 B-Cost(R5):96

Traffic to Network A

backup,and balancing using OSPF -Route propagation (under

Network A

87

Cost: 10

Copyright © 2000 Internet Initiative Japan Inc.

R1

l Two lines can be used for different purposes, and when a failure occurs, the remaining line can be used as backup for the faulty line.

Cost: 10 A-Cost(R1):20 B-Cost(R5):96

R6 Cost: 133 A-Cost(R5):153 Cost: 10 B-Cost(R4): 20

Sub circuit

2000/12/21

B-Cost(R4): 0 Network B A-Cost(R6):163

R4

l The backup is completed using the sub circuit.

A-Cost(R1): 0 B-Cost(R2):96 B-Cost(R5):96

– It needs other measures than OSPF configuration to backup with ISDN.

Cost: 10

R3

Cost: 66

A-Cost(R1): 20 B-Cost(R6):153

Cost: 10

l The sub circuit can't be cut off because OSPF HELLO packets flow in the backup lines as well.

Copyright © 2000 Internet Initiative Japan Inc.

A-Cost(R6): 163 B-Cost(R4): 20

R5 Network A

l Two lines can be used for different purposes, and when a failure occurs, the remaining line can be used as backup for the faulty line.

2000/12/21

R2

R1

– It needs other measures than OSPF configuration to backup with ISDN.

Copyright © 2000 Internet Initiative Japan Inc.

Main circuit Cost: 10 A-Cost(R1): 20 B-Cost(R6): 163

Cost: 10

l The sub circuit can't be cut off because OSPF HELLO packets flow in the backup lines as well.

2000/12/21

Backup using OSPF -Traffic flows (when a failure occurs)

A-Cost(R1):20 B-Cost(R2):96 B-Cost(R6):86

Cost: 66

B-Cost(R4): 0 Network A-Cost(R3):96 B A-Cost(R6):96

R4

R6 A-Cost(R3):96 A-Cost(R5):86 B-Cost(R4):20

Cost: 10

Cost: 10

Traffic to Network A

l Due to a failure, the network information between R2 and R3 is deleted.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

Cost value Router name of the propagation source (NEXT HOP)

89

l Using OSPF, respective lines can be balanced to use under normal conditions.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

Traffic to Network B

90

15

Backup and balancing using OSPF - Traffic flows (when a failure occurs) Cost: 10 A-Cost(R1):20 B-Cost(R5):96

Cost: 10

R2

A-Cost(R1): 0 B-Cost(R5):96

A-Cost(R6):96 B-Cost(R4):20

R3

Cost: 66

R1

B-Cost(R4): 0 Network A-Cost(R6):96 B

R5 A-Cost(R1):20 B-Cost(R6):86

R6

Cost: 66

A-Cost(R5):86 B-Cost(R4):20

Cost: 10

Cost: 10

Traffic to Network A

l The line which doesn ’t have the failure is used to backup.

2000/12/21

Traffic to Network B

Copyright © 2000 Internet Initiative Japan Inc.

91

OSPF settings for beginners -1

n

l When they are applied for a LAN, 100Mbps media can be used as 200Mbps media.

2000/12/21

n

OSPF has the concept called “Area” to aggregate routes. In a small -sized network, it can be constructed by setting the backbone Area as Area 0, and there are no needs to divide Areas for construction. Any Areas other than Area 0 always need to have contacts with Area 0. If the areas are indiscriminately divided, the expansion of the backbone will become difficult. BGP + OSPF is the mainstream of large-scale networks including ISP, and BGP has superiority in route aggregation. For those reasons, Areas except the backbone Area is used little.

– Always set a default route by “static”, and then inject default route by OSPF.

n

When a route injection from static and/or RIP other than OSPF, it affords to select either External Type 1 or External Type 2. What is External Type 1? —It

adds the OSPF cost from the point of the route injection to t he router which receives the OSPF route to the cost obtained at the time of injection to evaluate. When the same routes are injected, it is used to control choosing the closest interface. In the case of static, the point of the injection can be determined as the closest point, therefore, Type 1 is suitable.

n

What is External Type 2? —The

injected cost is maintained. When same routes are injected, evaluation is made based on the priority given at the time of the injection. This is effective to substantialize the BGP and other protocol ni formation by OSPF, however, it is not quite meaningful because BGP practically can ’t run on OSPF without any modifications. —Note: Cisco router’ s default setting is External Type 2. n

—Besides

OSPF costs, External Type 1 has priority over External Type 2. Therefore, switching at the time of a failure will become diffic ult.

Copyright © 2000 Internet Initiative Japan Inc.

93

l Router ID – No needs to concern about it in the case of small-sized networks but it is better to set the loopback interface. n n

n

n n

224.0.0.9 UDP 520

– OSPF n n

224.0.0.5/224.0.0.6 Protocol 89

l When Multicast is not supported – Some OS can ’t handle multicast. In this case, use broadcast as substitute.

OSPF gives the priority to DR (Designated Router), BDR (Backup DR), or DROTHER, or the start-up order. In the case of multimedia communications such as Ethernet, DR controls information. For those reasons, it better to start up with a router with higher performance to control information. In many small-sized networks, it is not necessary to concern. Copyright © 2000 Internet Initiative Japan Inc.

94

– RIP 2

– Better to start up with a router with higher performance and smaller load.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

l Check if a router’s filter restricts multicast addresses, protocols, and ports.

OSPF uses router ID (the IP address assigned for a router) for router to router communications. Normally, when the loopback interface is set up, its address will be used. When the identical address is assigned for the loopback interfaces of multiple routers, malfunction occurs. Attentions need to be paid.

l The order to start up routers n

2000/12/21

Trouble shooting - RIPv2 and OSPF don’ don’ t propagate

OSPF settings for beginners -3

n

Don’t mix External Type 1 and External Type 2

If it affords, use External Type 1.

2000/12/21

92

– Similar with a default route, use External Type 1.

l Default route n

Copyright © 2000 Internet Initiative Japan Inc.

OSPF settings for beginners -2

– Always set 0

n

l Two lines are effectively used to reduce line costs.

l Inject routes from static

l Area n

l When a failure occurs, 50% of the bandwidth is used for backup. l Balancing is basically achieved by the ratio of 1 to 1, therefore, it is difficult to balance the lines whose speeds are different.

R4

Network A

Characteristics of backup and balancing

95

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

96

16

Conclusions of dynamic routing

Fault-resistant network configuration Faultusing dynamic routing protocol

l Considering VLSM, the introduction of RIP 2 and/or OSPF is desired.

l Backup and balancing using the dual structure + OSPF

l For a simple network configuration, choose static.

l Backup by ring topology

l When only default routes are used, RIP is sufficient enough.

l ATM failure detection

l To implement balancing and others, use OSPF.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

97

Backup and balancing using the dual structure + OSPF - Connection diagram

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

98

Backup and balancing using the dual structure + OSPF‐ OSPF ‐Route propagation (under normal conditions) Network A

Network A

Network X switch

Network A

Network X

R

R

R

R

R

Network A

Switch

Network A

Network Y switch Network A

Switch

2000/12/21

l Use OSPF to advertise the Network A routing information.

Network Y

R

R

l The routing information equivalently propagates from 2 switches to respective routers.

Copyright © 2000 Internet Initiative Japan Inc.

99

Backup and balancing using the dual structure + OSPF -Route propagation (when a failure occurs) Network A

Failure R

R

Network X switch

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

100

Backup and balancing using the dual structure + OSPF - Traffic flows (when a failure occurs) Network X switch

Failure

R

R

R

R

Network A

Network A

Network Y switch

Network A

Network Y switch

Network A

l Due to a failure, the propagation of routing information partially changes.

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

l When a failure occurs, use either of those 2 switches to avoid the failure.

101

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

102

17

Backup by ring topology -Traffic flows (when a failure occurs)

Backup by ring topology - Route propagation (under normal conditions) Selected because its distance is the smallest

Not Selected because of its greater distance R

Network A

Network A Distance=1

R

R

Failure

Distance=2

R

R

R

Network A Distance=1

Network A

Network A

l Use RIP to advertise the Network A routing information.

l When a failure occurs, make a detour to back up communications.

l Under normal conditions, the shortest route has the priority. 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

103

ATM failure detection -1 Network A

Network B

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

ATM failure detection -2 Network B

Network A

Network A

R Network B

Network B

OSPF

ATM line

R

104

ATM line

R

R

Network A

l It can’t detect that VP is down to automatically shut down the interface (Cisco IOS11.X).

l When OSPF is used for dynamic routing to implement balancing, even an ATM line can detect a failure.

l For this reason, when the static routing is set up as described above to bundle 2 ATM lines, the desired backup can ’t be achieved. Failure Network B Network A Network B Network A

Failure

Network A

OSPF

Network B

ATM line

R

R

ATM line

R

l OSPF detects a failure, and stops using the line. Therefore, no packets will be lost.

R Network B

Network A

l In this case, approximately 50% of packets will be lost. 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

105

Backup and balancing technologies except dynamic routing

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

106

HSRP--1 HSRP Failure

l STP (spanning tree protocol) – Redundant structure in Layer 2 – When a failure occurs, it takes approximately 10 seconds to change the spanning tree.

R

R

R

R

l FDDI DAS (dual attachment station) – Redundant structure in Layer 2 – Almost instantly, it switches.

default Server

l I/F down and static

– Under default settings, the following shutdown occurs.

Tries to use the OAM cell as a substitute of“keepalive” in order to detect the line failure (IOS12.X).

n n

l HSRP

Copyright © 2000 Internet Initiative Japan Inc.

10 seconds for switching (recently, 3 seconds) 30 seconds for switching back (recently, 9 seconds)

– When routers are connected to switches, a discrepancy arises in the correspondence between ports and MAC addresses, and, in some cases, the switching will take more time.

– Instead of using dynamic routing at servers, one virtual MAC address is shared by multiple routers to implement switching when a failure occurs. 2000/12/21

Server

l When a failure occurs, the correspondence between MAC addresses and routers changes

– When it detects that an I/F down, the routing which directs the interface is deleted. This is the backup which uses this fact. – However, it can ’t be applied for ATM leased lines because line failure doesn ’t result in I/F down. n

default

107

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

108

18

HSRP--2 HSRP

HSRP--3 HSRP l HSRP+Interface Tracking (when a failure occurs)

Failure

R

R

l HSRP+Interface Tracking (under normal operation)

R

– Switches by Interface Tracking – Stops for 10 seconds (3 seconds, recently)

R

default Server

default Server

l HSRP+Interface Tracking (when a failure occurs)

l When a failure occurs, the faulty interface is detected, and it implements tracking to switch to active routers.

R

– Due to recovery, switching back occurs. – Stops for 30 seconds (10 seconds, recently). – Recent firmware provides the HSRP Delay function to eliminate the shutdown time derived from switching back.

R

default Server

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

109

HSRP--5 HSRP

2000/12/21

l Use multiple groups to apply MHSRP, the traffic will be separated for respective servers.

l HSRP+OSPF (Fault recovery phase)

OSPF R

– Switching back occurs due to fault-recovery. – No shutdown because only routes are switched.

default Server

R

l Dynamic routing doesn ’t associate switching back with shutdown, therefore, it ’s better to use dynamic routing such as OSPF and others for the router to router communications.

Copyright © 2000 Internet Initiative Japan Inc.

R

111

2000/12/21

Server

Copyright © 2000 Internet Initiative Japan Inc.

on the left Server

l Even if the size is small, the segment for the server is separate.

R

→ To assure the safety for the server

R

Server

l Clients obtain address allocation and default routes by DHCP. H

l However, MHSRP has the group ID conflict problem, therefore, attentions need to be paid when it is used for open networks.

Copyright © 2000 Internet Initiative Japan Inc.

112

Design to consider the future expansion -1 Features of the network configuration

default

2000/12/21

– Respective servers direct defaults to their corresponding HSRP virtual addresses.

default

l MHSRP (when a failure occurs)

Failure

l MHSRP (under normal operation)

R

Server

MHSRP--2 MHSRP

Server

110

MHSRP--1 MHSRP

R

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

113

2000/12/21

H

H

l It protects the server against the impact made by the broadcast flood.

Copyright © 2000 Internet Initiative Japan Inc.

114

19

Design to consider the future expansion -3 Add a server segment

Design to consider the future expansion -2 Add a server Server

Server

l Client segment broadcast can be confined to the segment, therefore, it prevents the broadcast flood phenomena from arising.

R

H

Server

H

H

Server

Server

l When more segments are added, it can be handled only by accelerating the speed of the backbone segment.

R

R Backbone segment 100BaseTXswitch, Giga bit Ethernet, FDDI switch

R

R

H

H

H

H

H

H

H

H

Add a network 2000/12/21

Add more networks

Copyright © 2000 Internet Initiative Japan Inc.

115

Network design

l Introduce switches for the servers and routers on which traffic concentrates.

l How can the addresses be allocated to respective departments? l How can the addresses be allocated to respective hosts in respective departments?

Address allocation in expectation of network expansion

117

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

Bad example 192.168.1.0/24

254

118

Address allocation for respective departments -1

Address allocation for the entire organization -1 1

116

l Use address from the beginning in ascending order or from the beginning and the end?

l Design the network topology, expecting future expansion.

Bad example

Copyright © 2000 Internet Initiative Japan Inc.

l Assuming the future expansion, the network addresses need to be allocated in the organization.

l Considering the safety, servers should be allocated in different segments.

Copyright © 2000 Internet Initiative Japan Inc.

2000/12/21

What is the address allocation policy?

l Considering scalability, creation of subnets is inevitable.

2000/12/21

l When a network expands from the switching-based network to the one described on the left, renumbering becomes inevitable.

H H

H

Server

l Add a server, while ensuring the safety of the server segment.

Divide into

1

192.168.1.0/24 9 10 19 20

Handle Department A as a subnet

subnets Server

Renumbering

Router

Department A

Department B

Department C

1415 16 17 192.168.1.0/25

192.168.1.128/25 Router

Department A

l When the addresses are used from the beginning and the end, renumbering becomes necessary when the network is divided into subnets. 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

Renumbering l When 10 addresses are

192.168.1.0/28

119

2000/12/21

allocated to respective departments, creation of subnets always requires renumbering.

Copyright © 2000 Internet Initiative Japan Inc.

120

20

Address allocation for respective departments -2 Good example 1

Address allocation in a department -1 Bad example

192.168.1.0/24 14 1516 17

Department A

Creates 303132 33

subnets

Department A Creates

1

7

Server

PC

14

subnets

Router

Department B Transferred without changes

subnetsB

subnetsA

l Renumbering can be avoided by allocating addresses for respective departments in expectation of future subnet creation. Copyright © 2000 Internet Initiative Japan Inc.

Department A

2000/12/21

121

Address allocation in a department -2 Department A

Good example 1

Divides into 14

Department B

l When the address space is determined to allocate addresses depending on objects, such as routers and server in a department, it can ’t support the newly created subnets, and renumbering becomes inevitable.

192.168.1.0/28

subnets

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

122

What is the address allocation policy ? l Use address from the beginning in ascending order or from the beginning and the end? – Use addresses from the beginning in ascending order.

routerServer PC

l How can the addresses be allocated to respective departments?

Server

– Consider subnets, and allocate 1 to 14 to the department A, and 17 to 30 to the department B, for example. Department A

l How can the addresses be allocated to respective hosts in respective departments?

Department B

l When the addresses are used from the beginning in ascending order, it can support newly created subnets without any obstacles. 2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

– Use addresses from the beginning in ascending order.

123

Conclusions -１ – A data link frame changes whenever interchange occurs. – IP datagram never changes. – The data link frame recipient doesn ’t always mean the IP datagram recipient.

l Differences between hubs and switches, as well as those between switches and routers – Allocate them effectively

l Routing is essential for connections with the Internet

124

l Use RIP 2 and/or OSPF to introduce VLSM. l Use dynamic routing to construct faultresistant networks. l Use OSPF to implement balancing and backup concurrently. l Allocate servers and others, for which the safety needs to be assured, to different segments. l Operate following the address allocation policy which concerns about the future expansion of the network.

l Once you understand the basic of dynamic routing, you can apply it Copyright © 2000 Internet Initiative Japan Inc.

Copyright © 2000 Internet Initiative Japan Inc.

Conclusions -２

l Difference between a data link layer and a network layer

2000/12/21

2000/12/21

125

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

126

21

［訳注： 1］unreachableと入力されていますが、パワーポイント上の表示がうまくい きません。ご確認をお願いします。 ［訳注： 2］原文はroutedとなっていますが、routerの間違いではないかと推測しま した。ご確認をお願い致します 。    いくつか全く同じページが含まれています。 P13と15、P14と16、P84と86、P85と87がそれぞれ同一のようです。 構成上の必要と推定して、そのまま翻訳しています。 ご確認をお願いします。

2000/12/21

Copyright © 2000 Internet Initiative Japan Inc.

127

22