TCP IP InterNetworking II

TCP/IP Internetworking IIChapter 9

Panko and PankoBusiness Data Networks and Security, 9th Edition© 2013 Pearson

Revised August 2013

Chapter (s)

Coverage Layers

1–4 Core concepts and principles All5 Single switched networks 1–26–7 Single wireless networks 1–2

8–9 Internets 3–410 Wide Area Networks 1-411 Applications 5

© 2013 Pearson 2

Perspective

Chapter 8◦ Major TCP/IP standards◦ Router operation

Chapter 9◦ Managing Internets◦ Securing Internets

© 2013 Pearson 3

Perspective

IP Subnetting

Network Address Translation (NAT)

DNS and DHCP

SNMP

Multiprotocol Label Switching

Securing Internet Transmission

IPv6 Management

© 2013 Pearson 4

Companies are given host parts by their ISP or an Internet number authority.

They divide the remaining bits between a subnet part and a host part.

Larger subnet parts mean more subnets, but this results in smaller host parts, which means fewer hosts per subnet.

The reverse is also true.

© 2013 Pearson 5

9.1: IPv4 Subnetting

If a part has N bits, it can represent 2N - 2 subnets or hosts per subnet.◦ 2N because if you have N bits, you can

represent 2N possibilities.◦ Minus 2 is because

you cannot have a part that is all zeros or all ones.

© 2013 Pearson 6


Part Size(bits) 2N 2N-2

4 24 = 16 16-2 = 148 ? ?

12 4,096 4,09465,536 65,53416

10 ? ?

© 2013 Pearson 7


DescriptionStep

32Total size of IP address(bits)1

Size of network partassigned to firm (bits)2 16

Remaining bits for firmto assign3 16

Selected subnet/host partsizes (bits)4 8 / 8

Number of possiblesubnets (2N - 2)

254(28 - 2)

Number of possible hostsper subnet (2N - 2)

254(28 - 2)

By definition

Assigned tothe firm

Bits for thefirm to assign

The firm’sdecision

© 2013 Pearson 8


DescriptionStep



Remaining bits for firm toassign3 16

Selected subnet/host partsizes (bits)4 6/10


62(26 - 2)


1,022(210 - 2)

By definition

Assigned tothe firm



© 2013 Pearson 9


DescriptionStep






4,094(212 - 2)


4,094(212 - 2)

By definition

Assigned tothe firm



© 2013 Pearson 10

9.1: IPv4 SubnettingDescriptionStep






254(28 - 2)


65,534(216 - 2)

By definition

Assigned tothe firm



© 2013 Pearson 11




Host part size (bits)4 ?

Number of possiblesubnets (2N - 2) ?

Number of possible hostsper subnet (2N - 2) ?

Selected subnet partsize (bits)Added 4

Exercise

Size of IP address2 32

© 2013 Pearson 12




Host part size (bits)4 ?

Number of possiblesubnets (2N - 2) ?

Number of possible hostsper subnet (2N - 2) ?

Selected subnet partsize (bits)Added 6

Exercise

Size of IP address2 32

IP SubnettingNetwork Address Translation (NAT)DNS and DHCP

SNMP



IPv6 Management

© 2013 Pearson 13

NAT◦ Sends false external source IP addresses and

port numbers that are different from internal source IP addresses and port numbers.

◦ For security purposes.◦ To have many more internal IP addresses than

your ISP gives you external IP addresses.

© 2013 Pearson 14

9.2: Network Address Translation (NAT)

© 2013 Pearson 15

9.3: NAT Operation NAT Firewall puts the real source IP address and port

number in the table.

© 2013 Pearson 16

9.3: NAT OperationNAT Firewall replaces the

source IP address and port number of the packet with a false source IP address and port

number.Adds to table.

© 2013 Pearson 17

9.3: NAT Operation NAT Firewall reverses the process for incoming packets.

NAT is Transparent to Internal and External Hosts.◦ The NAT firewall does all the work.◦ Neither host knows that NAT is taking place.◦ So there is no need to modify how hosts work.

© 2013 Pearson 18

9.2: Network Address Translation

Security Reasons for Using NAT◦ External attackers can put sniffers outside the

corporation.◦ Sniffers read IP addresses and port numbers.◦ Attackers can send attacks to these addresses

and port numbers.◦ With NAT, attackers learn only false

external IP addresses. Cannot use thisinformation to attack internal hosts.

© 2013 Pearson 19

9.2: Network Address Translation

Expanding the Number of Available IP Addresses◦ Companies may receive a limited number of IP

addresses from their ISPs.◦ There are roughly 4,000 possible ephemeral

port numbers for each client IP address.◦ So for each IP address, there can be up to about

4,000 external connections.◦ If a firm is given 248 IP addresses, there can be

roughly one million external connections.© 2013 Pearson 20

9.2: NAT

Expanding the Number of Available IP Addresses◦ If each internal device averages several

simultaneous external connections, each one will require a different port number.

◦ However, there should not be a problem with this many possible external IP addresses and port numbers.

© 2013 Pearson 21

9.2: NAT

Companies often use private IP addresses internally.

These can be used only within companies—never on the Internet.

There are three Private IP address ranges.◦ 10.x.x.x◦ 172.16.x.x through 172.31.x.x◦ 192.168.x.x (most popular)

© 2013 Pearson 22

9.2: NAT

There Are Protocol Problems Caused by NAT◦ IPsec, VoIP, and other applications have a

difficult time with NAT firewall traversal.◦ They must know the real IP address and port

number of the host on the other side of the NAT firewall.

◦ There are NAT firewall traversal techniques, but they must be managed carefully.

© 2013 Pearson 23

9.2: NAT

IP SubnettingNetwork Address Translation (NAT)DNS and DHCPSNMP



IPv6 Management

© 2013 Pearson 24

The Domain Name System (DNS)

© 2013 Pearson 25

© 2013 Pearson 26

9.4: Domain Name System (DNS) Lookup

Originating host needs the IP address of host dakine.pukanui.com.Asks its local DNS server at

Hawaii.edu.

© 2013 Pearson 28


Sends response to local DNS server,

not the client host.

© 2013 Pearson 29


Note that the local DNS server always sends back the response

message.

© 2013 Pearson 30

9.5: Domain Name System (DNS) Hierarchy

The DNS really is a general naming system for the Internet.

A domain is a set of resources under the control of an

organization.There is a hierarchy of domains.

© 2013 Pearson 31


The root is all domains.There are 13 DNS root servers.

© 2013 Pearson 32


There are two kinds of top-level domains.Generic top-level domains indicate organization

type (.com, .edu, .gov, etc.).Country top-level domains are specific to a

country (.UK, .CA, .CH, etc.).

Traditionally, generic top-level domains were strongly limited in number.

There have been a few additions over the year, such as .museum, .name, and .co.

As of 2013, any individual or company can propose to administer a generic top-level domain.


© 2013 Pearson 33

© 2013 Pearson 34


Companies want second-level domain names.(Microsoft.com, apple.com, panko.com, etc.).

Competition for these names is fierce.

© 2013 Pearson 35


Most companies divide their organizations into subdomains or subnets.

© 2013 Pearson 36


At the bottom of the hierarchy are individual hosts.

The Dynamic Host Configuration Protocol (DHCP)

© 2013 Pearson 37

© 2013 Pearson 38

9.6: Dynamic Host Configuration Protocol (DHCP) Service

Typical configuration information:◦ IP address for the DHCP client to use◦ The subnet mask for the client’s subnets◦ The IP address of the client’s default router◦ The IP addresses of the firm’s multiple DNS

servers

© 2013 Pearson 41


The two are often confused because both give a client PC an IP address.◦ DHCP gives a client PC its own dynamic IP

address.

◦ DNS gives a client PC the IP address of a host the client wishes to send packets to.

DNS versus DHCP

© 2013 Pearson 42


SNMPMultiprotocol Label Switching


IPv6 Management

© 2013 Pearson 43

Core Elements (from Chapter 4)◦ Manager program◦ Managed device◦ Agents (communicate with the manager on

behalf of the managed device)

© 2013 Pearson 44

9.7: Simple Network Management Protocol (SNMP)

ManagedDevices

AgentsManager

Core Elements (from Chapter 4)◦ Management information base (MIB).◦ Stores the retrieved information.◦ “MIB” can refer to either the database on the

manager or to the database schema.

© 2013 Pearson 45


Manager MIB

Messages◦ Commands (sent by a manager to an agent)

Get (to get information from the agent) Set (to tell the agent to change how the

managed devices is operating)◦ Responses (sent from agent to manager)

© 2013 Pearson 46

9.7: Simple Network Management Protocol (SNMP)2

Get or Set Command

Response

Messages◦ Traps (alarms sent by agents).◦ SNMP uses UDP at the transport layer to

minimize the burden on the network.

© 2013 Pearson 47


Trap

Set Commands◦ Dangerous if used by attackers.◦ Many firms disable Set to thwart such attacks.◦ However, they give up the ability to manage

remote resources without travel.◦ SNMPv1: community string shared by the

manager and all devices (poor).◦ SNMPv3: each manager–agent pair has a

different password (good).

© 2013 Pearson 48


Objects (Figure 9-8)◦ Specific pieces of information◦ Number of rows in the routing table◦ Number of discards caused by lack of resources

(indicates a need for an upgrade)

© 2013 Pearson 49


Objects are NOT managed devices!

Objects are specific pieces of data about a managed device.

Categories of Objects◦System objects (one set per managed device)

System name System description System contact person System uptime (since last reboot)

© 2013 Pearson 50


Categories of Objects◦ IP objects (one set per managed device)

Forwarding (for routers), Yes if forwarding (routing), No if not

Cause of resource limitations Number of rows in routing table Rows discarded because of lack of space Individual row data

© 2013 Pearson 51

9.8: SNMP Object Model

Categories of Objects◦TCP objects (one set per managed device)

Retransmission time Maximum number of TCP connections allowed Opens/failed connections/resets Segments sent Segments retransmitted Errors in incoming segments Data on individual connections (sockets, states)

© 2013 Pearson 52


Categories of Objects◦UDP objects (one set per host)

Traffic statistics

◦ ICMP objects (one set per host) Number of ICMP errors of various types

© 2013 Pearson 53


Categories of Objects◦ One set per managed device:

System IP TCP UDP ICMP

Interface objects: one set per interface (port)

© 2013 Pearson 54


Categories of Objects◦ Interface objects (one set per interface)

Type (e.g., 69 is 100Base-FX; 71 is 802.11) Status: up/down/testing Speed Errors: discards, unknown protocols, and so on

© 2013 Pearson 55


SNMP Manager program collects data.◦ Places it in the MIB.

Visualization Program.◦ The administrator’s interface to the MIB.◦ Helps the administrator visualize patterns in the

MIB data.◦ Can order the SNMP Manager to collect certain

data or to send set commands to change the configurations of managed devices.

© 2013 Pearson 56

9.7: SNMP

User Functionality◦ Reports, diagnostics tools, and so on, are very

important.◦ They are not built into the standard.◦ They are added by network visualization

program vendors.◦ Critical in selection of a network management

vendor.

© 2013 Pearson 57



SNMPMultiprotocol Label SwitchingSecuring Internet Transmission

IPv6 Management

© 2013 Pearson 58

Routers route each packet individually, going through the three steps we saw in the last chapter.◦ Even if the next packet is going to the same

destination IP address, the router will go through all three steps.

◦ This consumes a great deal of processing power per packet.

◦ This makes traditional routing expensive.

© 2013 Pearson 59


MPLS addresses this issue.◦ Routers identify the best route for a range of IP

addresses before sending data.◦ That route is given a label number.◦ Each packet in a stream gets a label with this

label number.◦ Routers do only a quick table lookup per packet.◦ Table lookups require little processing power.◦ So multiprotocol label switching is much less

expensive than traditional routing.© 2013 Pearson 60


© 2013 Pearson 61

9.9: Multiprotocol Label Switching (MPLS)

© 2013 Pearson 62


Label Number is 123

Label sits between the frame header and the IP packet header.

9.9: MPLS

© 2013 Pearson 63

IP PacketHeader MPLS Label Frame Header

© 2013 Pearson 64


Router 3 sends the packet out

through Interface 1

Implementing MPLS is difficult. Many individual ISPs and corporations do

it. Some individual ISPs have “peering”

arrangements with other individual ISPs to do it.

There is no general way to move MPLS out to all ISPs and organizations.


© 2013 Pearson 67


SNMP

Multiprotocol Label SwitchingSecuring Internet TransmissionIPv6 Management

© 2013 Pearson 68

Security was not addressed in the initial design of TCP/IP.

Jon Postel, who edited the main Internet RFCs, explained to the first author, “It just wasn’t a problem then, and we were stretched thin.”

Today, firms are adding security to their transmissions through IPsec VPNs.


© 2013 Pearson 69

A virtual private network (VPN) is a cryptographically secured transmission path through an untrusted environment.◦ The Internet◦ A wireless network◦ Communication in a foreign country

Like having your own private network in terms of security.◦ However, not a real private network.

9.10: Virtual Private Network

© 2013 Pearson 70

IPsec VPNs

© 2013 Pearson 71

9.10: IPsec VPNs

© 2013 Pearson 72

There are two types of VPN:

Remote access VPNs connect a remote user

to a corporate site.The user connects to a

VPN gateway at the site.

9.10: IPsec VPNs

© 2013 Pearson 73

There are two types of VPNs:Site-to-site VPNs protect all traffic

traveling between two sites.Each site has a gateway to encrypt

outgoing traffic and decrypt incoming traffic.

IPsec has two modes (ways) of operating:◦ Transport mode◦ Tunnel mode

Each mode has strengths and weaknesses. Selecting an IPsec mode option is very

important to security.

9.11: IPsec in Transport and Tunnel Modes

© 2013 Pearson 74

© 2013 Pearson 75

9.11: IPsec Transport and Tunnel Modes

In transport mode, IPsec provides protection over the Internet and also over site networks between the

hosts.

© 2013 Pearson 76


Transport mode requires a digital certificate and configuration work on each host.

This is expensive.

© 2013 Pearson 77


In tunnel mode, IPsec only provides protection over

the Dangerous Internet—not within site networks.

© 2013 Pearson 78


Only the two IPsec gateways need digital

certificates and configuration work.

Criterion Transport Mode Tunnel ModeSecurity Better because it

provides host-to-host protection.But firewalls cannot read encrypted traffic.

Not as good because it only provides security over the Internet or another trusted network (a wireless network, etc.).

Cost Higher because of configuration work on each host.

Lower because IPsec operates only on the IPsec gateway .

© 2013 Pearson 79


© 2013 Pearson 80

9.12: IPsec Security Associations and Policy Servers

© 2013 Pearson 81

9.12: IPsec Security Associations and Policy Servers

SSL/TLS VPNs

82© 2013 Pearson

Purpose◦ To provide a secure connection between a client

browser and a webserver application on a webserver host

◦ Use is indicated by https:// in the URL◦ Very widely used

9.13: SSL/TLS VPNs (Study Figure)

© 2013 Pearson 83

Origin◦ Created by Netscape as SSL.◦ IETF took over the standard.◦ IETF changed the standard’s name to Transport

Layer Security (TLS).◦ We refer to the standard, generically, as

SSL/TLS.


© 2013 Pearson 84

Attraction of SSL/TLS◦ Universally supported by browsers and

webserver applications.◦ So no added cost on the client to use it!◦ No extra software on the server is needed, but

SSL/TLS must be configured, which usually is simple.


© 2013 Pearson 85

Limitations of SSL/TLS◦ Operates at transport layer so no protection for

IP or transport headers◦ Limited to applications written to work with

SSL/TLS: HTTP and e-mail, primarily◦ Cryptographically weaker than IPsec

Has been partially cracked◦ No policy servers for centralized management


© 2013 Pearson 86

Overall◦ Decent quality, cheap, and easy security◦ Limited in how it can be used and managed

Comparison with IPsec◦ IPsec is more complex and so more expensive.◦ Can be used for all types of VPNs.◦ Can be managed well.◦ Gold standard in TCP/IP security.


© 2013 Pearson 87

Transition from IPv4 to IPv6 IPv6 subnetting IPv6 configuration Other IPv6 standards

◦ ICMPv6◦ Extending DNS◦ Replacing the Address Resolution Protocol

IPv6 Management Issues

© 2013 Pearson 89

Must transition all clients, routers, firewalls, and so on

The IETF’s plan◦ No backward compatibility◦ Instead, add both IPv4 and IPv6 protocol stacks

at the internet layer to all new devices◦ As soon as most devices have IPv6 protocol

stacks, configure the devices and add IPv6 support to IPv4 support

◦ Eventually, turn off IPv4 support

Transitioning to IPv6

© 2013 Pearson 90

Problems and reactions◦ IPv6 offered few benefits, so most companies

ignored IPv6.◦ The shortage of IPv4 addresses was handled

(intelligently) through NAT.◦ But now, IPv4 addresses are gone.◦ Now some clients, such as mobile phones, only

have IPv6 stacks at the protocol layer.◦ To serve them, companies are rushing to turn

on and configure IPv6 support.

Transitioning to IPv6

© 2013 Pearson 91

Must deal with global IPv6 unicast addresses◦ Like public IPv6 addresses◦ Have 3 parts but different names

Subnetting

© 2013 Pearson 92

IPv6 Address Part

Corresponding IPv4 Address Part

Length of IPv6 part

Routing Prefix Network Part VariableSubnet ID Subnet Part VariableInterface ID Host Part 64 bitsTotal 32 bits 128 bits

9.15: Global Unicast Addresses in IPv6

© 2013 Pearson 93

Global Routing Prefix(network part in IPv4)

Subnet ID(subnet part

in IPv4)Interface ID

(host part in IPv4)


© 2013 Pearson 94



in IPv4)Interface ID

(host part in IPv4)

(Almost)Always 64 bits

Interface ID is not of variable length like IPv4 host parts.

“Waste” 64 bits, but have plenty to lose.

An IP address registrar gives you a 32-bit global routing prefix.

How long is your subnet ID? How many subnets can you have

(approximately)? Many companies have a two-layer

hierarchy of subnets, using some bits for the main subnet and remaining bits for sub-subnets.

9.15: Subnetting

© 2013 Pearson 96

Modified 64-bit Extended Unique Identifier (EUI) Format

First, display the MAC address in hexadecimal notation (48 bits)◦ Remove dashes◦ Convert text

to lower case

9.16: 64-Bit Unicast Interface ID

© 2013 Pearson 97

AD-B1-C2-D3-E5-F5

adb1c2d3e5f5

Second, divide the address in half Insert fffe in the middle This creates a 64-bit address


© 2013 Pearson 98

adb1c2 fffe d3e5f5

adb1c2fffed3e5f5

Third, in the second nibble (d) (1101) Invert the second bit from the right (1111)

(f) This is called Modified 64-bit EUI


© 2013 Pearson 99

adb1c2fffed3e5f5

afb1c2fffed3e5f5

1. Begin with MAC in hexadecimal notation 2. Divide the 48 bits into 2 halves of 24 bits 3. Insert fffe between the two halves 4. Place into four-hex groups separated by

colons 5. Flip the second-least significant bit in the

first octet

9.16: 64-Bit Interface ID Recap

© 2013 Pearson 100

Hosts must be configured with IP addresses

IPv4 uses DHCP IPv6 offers two configuration mechanisms

◦ DHCPv6 (very similar to IPv4)◦ Stateless autoconfiguration, which does not use

a DHCPv6 server◦ Not available in IPv4

9.17: IPv6 Stateless Autoconfiguration

© 2013 Pearson 101

Stateless Autoconfiguration◦ The client configures itself, without using a

DHCPv6 server.◦ First, the client creates a link-local IPv6 address.◦ Second, the client creates a global unicast IPv6

address.


© 2013 Pearson 102

Creating the Link-Local IPv6 Addresses◦ Link-local IPv6 addresses can be used only

within a single network (wireless or switched wired).

◦ If the client does not need a global IP address, the autoconfiguration process can stop here.


© 2013 Pearson 103

Creating the Link-Local Address◦ First create a 64-bit interface ID using the MAC

address of the client.◦ Add a routing prefix 111 1110 10 followed by 56

bits of zeroes.◦ This is the link-local IP address: fe80::x, where x

is the octets of the EUI-64.


© 2013 Pearson 104

Testing the Link-Local Address◦ Another host may be using this address.◦ So the client uses the ICMPv6 neighbor

discovery protocol to ask if any other host in the single network is using this address.

◦ If none reply, the client may use this address within its single network.


© 2013 Pearson 105

Creating the Global Unicast IPv6 Address◦ Needed for communication over the Internet.◦ Begin with the link-local address.◦ Keep the interface ID but get a new routing

prefix and subnet ID.◦ Client sends an ICMPv6 router solicitation

message to the address FFF02::1, which all routers listen for.


© 2013 Pearson 106

Creating the Global Unicast IPv6 Address◦ Routers respond with IPv6 router advertisement

messages.◦ The router advertisement message may state

that autoconfiguration is not allowed.◦ If this is not the case, the message gives the

routing prefix and subnet ID.◦ The client now has a global unicast IPv6

address.


© 2013 Pearson 107

Limits◦ More limited than traditional DHCP

autoconfiguration.◦ At a minimum, router advertisement messages

give only a routing prefix and subnet ID.◦ Of course, the packet containing the router

advertisement message gives the IPv6 address of the router, which becomes the default router.


© 2013 Pearson 108

Uses◦ How can a client get other IPv6 configuration

information?◦ If a client is a dual-stack client, the IPv4 stack

can obtain full configuration information, which the IPv6 stack can use.


© 2013 Pearson 109

Uses◦ If the client is not a dual-stack client, it needs at

least one more piece of configuration information—the IPv6 addresses of DNS servers.

◦ The IETF has extended router advertisement messages to provide the IPv6 addresses of DNS servers.

◦ However, this is only an option.


© 2013 Pearson 110

Known Security Weaknesses◦ An attacker might create an address that does

not use its proper EUI-64.◦ An attacker may create an address that uses

the EUI-64 of another host to impersonate it.◦ Several operations can be used to create

flooding denial-of-service attacks.


© 2013 Pearson 111

IPv6 Address Renumbering◦ Stateless autoconfiguration may be used to

renumber all IP addresses in a firm automatically, changing subnet IDs and even routing prefixes.


© 2013 Pearson 112

ICMPv6◦ Many new types were created for neighbor

discovery, stateless autoconfiguration, and so on.

9.18: Other IPv6 Standards

© 2013 Pearson 113

Domain Name System (DNS)◦ The DNS information for a host is contained in

several records.◦ DNS A Record. The A record contains the IPv4

address for the target host.◦ DNS AAAA Record. For IPv6 addresses, a new

address record had to be added. IPv6 addresses are four times as long as IPv4

addresses, so the added record is called the AAAA record.


© 2013 Pearson 114

Address Resolution Protocol (ARP) Messages◦ In IPv6, handled by the ICMP neighbor discovery

protocol, which has two message types.◦ Neighbor solicitation messages ask host to

respond.◦ Neighbor advertisement messages give the

host’s data link address.◦ There is no ARPv6.


© 2013 Pearson 115

IP SubnettingNetwork Address Translation (NAT)DNS and DHCPSNMPMultiprotocol Label SwitchingSecuring Internet TransmissionIPv6 Management

© 2013 Pearson 116

Where We’ve Been

Chs. Title Layers1-4 Core Concepts All5-7 Single Networks 1 and 28-9 Internets 3 and 410 Wide Area Networks 1-411 Networked

Applications5

Where We are Going

© 2013 Pearson 117

TCP IP InterNetworking II

Documents

Transcript of TCP IP InterNetworking II