IPv4 Multicasting and IPv6 Introduction

Question 1

Multicasting

Answer 1

Packets routed from source to multiple destinations

• Key for group communication
• Address identifies a group Not widely used for now.

Question 2

IPv4 Multicast Addressing

Answer 2

Class D addresses

• Begin with 111O → 224.O.O.O - 239.255.255.255
• Group delivery delegated to lower level (MAC)
• IP multicast address mapped to a MAC multicast address: 0x01005E U 0x1b U 23 LSBs of IP address
• Interface card configured to receive that MAC multicast
• Recipient initiated group join

Question 3

Routers in IPv4 Multicasting

Answer 3

Routers role:

• Routers discover host groups on each LAN
  
  • Internet Group Management Protocol (IGMP)
• Routers announce host groups to others
  
  • Multicast routing protocols
• Routers build a distribution tree foreach host group
  
  • To all LANs with at least a member

Question 5

State of deployment of IPv4 Multicast

Answer 5

• Not widely supported
• Not fit to common traffic engineering practices
• Mostly limited to controlled environments

Question 6

Why IPv6?

Answer 6

• Larger address space
• More effiecient on LANs
• Multicast and anycast
• Security
• Policy routing
• Plug and Play
• Traffic differetiation
• Mobility
• Quality of service support

Question 7

IPv4 usable addresses and why we need more of them

Answer 7

Only 3.5M addresses can be used

Those 3.5M can be used hierearchically

• The prefix used in a physical network cannot be used into another
• Lots of unused addresses

Question 8

Interim IPv4 solutions to the saturation of address space

Answer 8

• Taylored sized network => NETMASK
• Private addresses
  
  • Intranet, however not enough. It should be used in conjuction with NAT or ALG.
• NAT
  
  • Very popular
  • Proposal for RSIP
• ALG: Application Level Gateway

Question 9

Routing scalability issues of IPv4

Answer 9

• Routing table size
  
  • Internet size
  • each subnetwork must be advertised
• Problems
  
  • Router resource limitations
  • Too much info to manage
  • Routing protocols limitations
  • High probability of route changes (for backbone routers)

Question 10

Routing scalability solutions in IPv4

Answer 10

Aggregate multiple routes into one:

• Short prefix including others
• CIDR (Classless inter-domain routing)
• Limited by non-rational assignment of IP prefixes

Limit the assignement of IP addresses:

• Regional Internet Registry: assign address blocks only to big players.

Scalability of routing protocols:

• No solution, at present. Still open also for IPv6

Question 11

Addressing efficiency

Answer 11

H = log10(number of adresses) / number of bits

Question 12

Addressing space organization of IPv6

Answer 12

Question 13

Address structure IPv6

Answer 13

With subnet prefixes having no more than 64bits

Question 14

Subnet ipv6

Answer 14

set of hosts with same prefix

Question 15

Link ipv6

Answer 15

Physical network

subnetwork === link

Question 16

On-link hosts

Answer 16

Have same prefix and communicate directly

Question 17

Off-link station

Answer 17

Different prefixes → communicate through router

Question 18

Link local

Answer 18

frist bits are 1111 1110 10

Si riferiscono agli indirizzi privati “automatici”, generati dall’autoconfigurazione, che è il processo in cui una stazione genera automaticamente un indirizzo per connettersia un link IPv6.

Question 19

Site local

Answer 19

Is deprecated function that is equivalent to ipv4 private addresses.

frist bits are 1111 1110 11

Question 20

Global Unicast Aggregatable

Answer 20

• Equivalent to public IPv4 addresses.
• They start with 3 bits: 001
• Topology-based assignement:
  
  • Service provider hierarchy
  • Effective aggregation

Question 21

Global Univast Addressing Space Organization

Answer 21

IPv4 Mapped addresses

• 80 bits to 0, 16 bits to 1 then the IPv4 address: 0:0:0:0:0:ffff:A00:1
• To be used during transition phase

IPv4 compatible addresses

• 96 bits to 0 then IPv4 address: 0:0:0:0:0:0:A00:1
• Spacial notation: ::10.0.0.1
• Deprecated

Question 22

Plug And Play

Answer 22

Scenarios:

• Dentist office
• Thousand computers on the dock

Solution: autoconfiguration:

• Stateless: no server needed
• Stateful: DHCP server

Question 23

IPv6 Special addresses

Answer 23

• Loopback: ::1
• All nodes (multicast) FF02::1
• All routers (multicast) address FF02:2
• Unspecified address: ::

Question 24

Changed and upgraded protocols in IPv6

Answer 24

Changed:

• IP
• ICMP
• ARP: Integrated in ICMP
• IGMP: integrated in ICMP

Upgraded:

• DNS
• RIP and OSPF
• BGP and IDRP
• TCP and UDP
• Socket interface

Question 25

Socket API

Answer 25

* Programming interface for TCP/IP services * Used in application implementation (UDP messages and byts on TCP connections) * Socket descriptor is equivalent to a file descriptor (pointer) for network resources * Associated to UPD/TCP session/connection * Server listens (listen()) for requests on a port and accepts requests * Client connects to port of remote server, send/receive data.

Question 26

IPv6 Packet header format

Answer 26

* Many fields removed respsect to IPv6 * Not useful, no longer needed, not used in each packet

Question 27

Extension Header Format

Answer 27

Extension headers: * used when useful, not needlessly processed in each packet * Hop by hop option: options that each router has to evaluate * destination option: options only for destination * Type-Length-Value Format Extension header must be 64bits aligned: * Pad1 Option: Type:0 * PadN Option: Type:1 Len:0-6 Value: 0-6 bytes

Question 28

Routing HEADER

Answer 28

* Used to list nodes to travers on the path to destination (loose or strict) * Segment Left Field show the number of remaining path segments

Question 29

IPv6 Encapsulation

Answer 29

Encapsulated in layer 2 frames (EtherType: 86DD): new protocol that enables dual stack (IPv4 as is + IPv6)

Question 30

IPv6 address mapping to MAC

Answer 30

IP unicast address: * Procedural: protocol based (Neighbor discovery) IP multicast address: * algorithmic mapping IPv6 multicast trasmission: * Based on MAC multicast * MAC A.: 33-33 | 4 LSBytes of IPv6 Address * Example: FFOC::89:AABB:CCDD → 33:33:AA:BB:CC:DD

Question 31

Neighbor discovery

Answer 31

New function in ICMP that replaces ARP, based on multicast, but most likely only one station gets involved. Host Cache === ARP cache Process 1. FF02::1:FF/104 | LSbits of IPaddr (Mulicast address) is solicited through Neighbor solicitation 2. Encapsuted in MAC frame to 33:33:FF | LSbits of IPaddr 3. Likely is just 1 host per group 4. ICMP Neighbor Advertisement to petitioner

Question 32

Transition to IPv6

Answer 32

Isolated IPv6 networks -\> islands grow -\> ipv6 native connectivity -\> doomsday All protocols specified since 1996 Implemented on routers, maybe less stable than ipv4. Hardware implementation on Layer 3 switch. Implemented on Windows since 2000, also on GNU/Unix ecosystem. When will it happen? When there will not be anymore addresses, but the issue is being mitigated (provident address assignment, use of private addressing, NAT, proxying) Acceptable limitation so far: - user traceability - Not many public addresses for servers - NAT not suitable for all applications