4.3

IP Internet Protocol

 

The Internet network layer

- host, router network layer functions : physical layer -> link layer -> network layer ( routing protocols, IP protocol, ICMP protocol ) -> transport layer (TCP, UDP)

 

IP datagram format

- IP header : 5 x 32bits ( 5 x 4bytes  = 20Bytes + options )

- IP payload : data

 

Header component

--

-- ver : IP protocol version number [ 4 bits ]

-- head.len : Header length [ 4 bits ]

-- type of service : "type" of datagram

-- length : total datagram length (bytes) [ 16bits ]

--

-- 16-bit identifier

-- flgs

-- fragment offset

--

-- TTL : max number remaining hops ( decremented at each router)

-- Protocol : upper layer protocol to deliver payload to 

 

How much overhead?

 - 20 bytes of TCP (header)

 - 20 bytes of IP (header)

 - = 40 bytes + app layer overhead

 

A Few notes

- TTL has many purposes, but the most important one is to ensure that datagrams do not circulate forever

- Protocol is what connects network layer to transport layer

    - ICMP = 1, TCP = 6, UDP = 17, IPv4 = 4, IPv6 = 41, etc

- Length is 16 bits, includes header + payload

    - but IP packets are rarely over 1500 Bytes, why? : L2 Ethernet MTU (Maximun Transport Unit size) : 1500 Bytes, WiFi :2300 Bytes

- Fragmentation (IPv4)

    - IPv6 does not allow fragmentation, why? later...

- Internet checksum is for error checking in IP header

    - must be re-computed at every router, why? -> TTL change

    - IPv6 does not have checksum, why? later..

- Options

    - IPv6 does not include options. why? later...

 

 

IP fragmentation, reassembly

- Network links have MTU

    - (max. transfer size) - largest possible link-level frame

    - different link types, different MTUs

- Large IP datagram divided ("fragmented") within net

    - one datagram becomes serveral datagrams

    - "reassembled" only at  final destination

    - IP header bits used to identifiy, order related fragments

- Examples ) 4000 byte datagram , MTU = 150bytes

  -> 4000 large datagram becones serveral smaller datagrams 1500, 1500, 1040 ( header 40 ) fragments

  -> fragflag -> last : 0 other : 1, offset -> first : 0, other : datalength / 8

- IPv6 does not allow fragmentation, why? : for speed, for Gbps + don't allow large datagram

 

IPv4 addressing : introduction

- IP address : 32 bit identifier for host router Interface.

    - Must be globally unique ( public IP address )

    - Usually written in dotted-decimal form : <8bit>.<8bit>.<8bit>.<8bit>

- Interface : connectio between host/router and physical link

    - router's typically have multiple interfaces

    - host typically has one or two interfaces ( wired Ethernet, wireless 802.11 )

- IP address associated with each interface!

- Q : How are interfaces actually connected ?

    A : Wired Ethernet interfaces connected by Ethernet switches ( aka. hub ),

          Wireless WiFi interfaces connected by WiFi AP

 

Subnets

- IP address

    - subnet part - high order bits

    - host part - low order bits

- What's a subnet?

    - device interfaces with same subnet part of IP address

    - can physically reach each other without intervening router

    - to determine the subnets : detach each interface from its host or router, creating islands of isolated networks

+ each isolated network is called a subnet

- For "x.x.x.x/Y", /Y is the number of bits used for subnet address

 

IP addressing : CIDR

- CIDR : Classless InterDomain Routing

    - subnet portion of address of arbitrary length

    - address format : a.b.c.d/x, where x is # bits in subnet portion of address

- Classful addressing?

    - There used to be classes.... : Subnet portion of address fixed.

 

IP addresses : How to get one?

- Q : How does a host get IP address?

    - hard-coded by system admin in a file

    - dynamically get address from as server -> DHCP

DHCP : Dynamic Host Configuration Protocol

- Goal : allow host to dynamically obtain its IP address from network server when it joins network( "plug-and-play")

      -> can renew its lease on address in use

      -> allows reuse of addresses (only hold address while connected/on)

      -> support for mobile users who want to join network ( more later )

DHCP overview

      -> host broadcasts "DHCP discover" msg [optional]

      -> DHCP server responds with "DHCP offer" msg [optional]

      -> Host requests IP address : "DHCP request" msg

      -> DHCP server sends address : "DHCP ack" msg

 

DHCP : more than IP addresses

- DHCP can return more than just  allocated IP address on subnet

  -> address of first-hop router (aka gateway) for client

  -> name and IP address of DNS server

  -> network mask (indicating netwokr versus host portion of address )

Example

-> Connecting laptop needs its IP address, addr of first-hop router, addr of DNS server -> use DHCP

-> DHCP request encapsulated in UDP, encapsulated in IP encapsulated in 802.1 Ethernet

-> Ethernet frame broadcast (dest : FFFFFFFFFFFF) on LAN, received at router running DHCP server

-> Ethernet demuxed to IP demuxed, UDP demuxed to DHCP

-> DHCP server formulates DHCP ACK containing client's IP address, IP address of first-hop router for client, mane & IP address of DNS server

-> encapsulation of DHCP server, frame forwarded to client, demuxing up to DHCP at client

-> client now knows its IP address, name and IP address of DNS server, IP address of its first-hop router

 

 

IP addresses : How to get one? continue

- Q : how does network get subnet part of IP addr?

    - A : gets allocated portion of its proviedr ISP's address space

      -> ISP's block ( subnet ) and allocated block to organization

- Q : how does an ISP get block of addresses?

    - A : ICANN : Internet Corporation for Assigned Names and Numbers

 

Hierarchical addressing : route aggregation

- Hierarchical addressing allows efficient advertisement of routing information : Out of ISP know just one IP address of Hierarchical addressing ( forwarding table longest prefix match )

 

NAT : network address translation

Translate one public IP address to private  IP address (home network)

- All datagrams leaving local netwokr have some single source : NAT IP address : 138.76.29.7, different source port numbers

- datagrams with source or destination in this network have 10.0.0/24 address for source, destination (as usual)

Motivation

 - loca lnetwork uses just one IP address as far as outside world is concerned

   -> range of addresses not needed from ISP : just one IP address for all devices

   -> can change address of devices in local network without notifying outisde world

   -> can change ISP without changing addresses of devices in local network

   -> devices inside local net not explicityly addressable/visible by outside work(security, side effect)

Implementation : NAT router must

 - outgoing datagrams (in to out)

   -> replace (src IP addr, port #) of every outgoing datagram to (NAT IP addr, new port #)

   -> remote machine will respond using (NAT IP addr, new port #) as destination addr

- remember (in NAT translation table)

   -> every (src IP addr, port #) to (NAT IP addr, new port #) translation pair

- incomming datagrams

   -> replace (NAT IP addr, new port #) in dest fields of every incoming datagram with corresponding (src IP addr, port #) stored in NAT table

 

- 16bit port-number field -> 60,000 simultaneous connections with a single LAN-side address!

- NAT is controversial

  -> routers should only process up to layer 3....

  -> address shortage should be solved by IPv6

  -> violates end-to-end argument : NAT possibility must be taken into account by app designers...

  -> NAT traversal : what if client wants to connect to server behind NAT?

 

IPv6

Motivation : Initial motivation : 32-bit address space soon to be completely allocated.

- Additional motivation : header format helps speed processing/forwarding + header changes to facilitate QoS

- IPv6 datagram format : fixed-length 40 byte header + no fragmentation allowed + no IP checksum

 

IPv6 datagram format

- Priority : identify priority among datagrams in flow

- flow label : identify datagrams in same "flow"

- next header : identify upper layer protocol for data

Other changes from IPv4

- checksum : removed entirely to reduce processing time at each hop

- options : allowed, but outside of header, indicated by "Next Header" field

- ICMPv6 : new version of ICMP -> additional message types + multicast group management functions

 

Transition from IPv4 to IPv6

- not all routers can be upgraded simultaneously

 -> no "flag days"

 -> how will network operate with mixed IPv4 and IPv4 routers?

- Tunneling : IPv6 datagram carried as payload in IPv4 datagram among IPv4 routers ( IPv4 datagram 에 IPv6 datagram 을 넣음. IPv4 헤더에 IPv6 가 있다고 알려줌)

 

IPv6 : adoption

- Google : 8% of clients access services via IPv6

- NIST : 1/3 of all US government domains are IPv6 capable

- Long time for deployment, use : 20 years and counting + tink of application-level changes in last 20 years : WWW, Facebook, streaming media, Skype, ...

 

Generalized Forwarding and SDN

Each router contains a flow table that is computed an distributed by a logically centralized routing controller