Network Layer: Ip

The IP protocol itself, Routing protocols, he Internet Control Message Protocol (ICMP)

Internet Control Message Protocol (ICMP), which is a network-layer protocol for error and status reporting.

IP component. a network-layer protocol for error and status reporting.

IP component. determines networkconnectivity and how forwardingtables are configured at routers

IP component. deals with addressinghosts, formatting datagrams, fragmenting and reassemblingdatagrams, and forwarding datagrams through routers.

a 32-bitsource IP address, which identifies the sender, and a 32-bitdestination IP address, which identifies the recipient.

a 20-byte header, a variable-sizeoptions field after the header, and the payload, which willtypically be the TCP or UDP segment.

time-to-live (TTL) field

a counter that isdesigned to keep packets from circulating indefinitely inthe network case forwarding tables accidentally create cycles.An IP datagram is typically initialized with a TTL of 60 or 64and the TTL is decremented by one each time it enters a router.If the TTL reaches zero, the router will discard the packet.

60 - 64

An IP datagram is typically initialized with a TTL of 60 or 64and the TTL is decremented by one each time it enters a router.

If the TTL reaches zero, the router will discard the packet.

the higher-layerprotocol that is contained within the data. Common values are6 to identify the data as a TCP segment and 17 to identify thedata as a UDP segment.

Common values are6 to identify the data as a TCP segment and 17 to identify thedata as a UDP segment.

a 16-bit header checksum.This is calculated with the same formula as UDP and TCP checksums.Only the IP header is checksummed. A router has to recomputethe checksum since the TTL field (and possibly the options field)will change with each network hop.

onl the IP header. the router has to recompute the Checksum because the TTL changes each hop.

because the TTL changes each time it hops. So the router must recompute the full checksum after it arrives.

If a packet is bigger thanthe MTU of the outgoing link, a router can split the datagraminto two or more fragments. Each fragment is a separate IPdatagram with its own IP header.

reassemble them, before passing them to the transport layer

a router can split the datagraminto two or more fragments. Each fragment is a separate IPdatagram with its own IP header.

two data fields and two one-bit flags. (Don't fragment (DF) flag and more fragments (MF) flag. ID number field in the IP header, 13-bit fragment offset.

A don’t fragment (DF) bit tells arouter that fragmentation is not permitted on a datagram. Thismay result in the inability to route the datagram.

When a datagram is fragmented, the IP header of each datagram holdinga fragment contains the same ID number. This tells the receiverthat those datagrams are part of the same original datagram.

Each fragment also contains a 13-bit fragment offset. This is anumber that is multiplied by eight to indicate where thedata in this fragment belongs in the reassembled datagram. Thefirst datagram contains an offset of zero.

zero

Each datagram fragmentexcept for the last one has a more fragments (MF)bit set to one. The last fragment will have MF=0 andthe fragment offset along with the IP length field will indicatethe length of the final reassembled datagram.

Every interface on an IP network must have a uniqueIP address. If a host has two interfaces (e.g., Ethernet and802.11 links), it will have one IP address for each link.If a router has 128 ports, it will have 128 IP addresses.

each router would haveto have to be able to look up an individual address ina forwarding table of over four billion addresses. Moreover,routing algorithms would need to manage information aboutthe route of every single address on the Internet.Instead, groups of adjacent addresses are assigned toan organization.

ability to use one prefix to refer to a route thatmay span multiple sub-networks or hosts

(also called asubnetwork or a network) is a group of adjacentIP addresses that share a common prefix and are assignedto an organization. A subnet makes up a logicalnetwork that is connected to a router.

Subnets are expressed in CIDR (Classless Inter-DomainRouting) notation, whose format is a 32-bit IP address thatcomprises the identifying bits of the subnetwork followed bya slash an the number of bits that identify the subnetwork.

128.6.0.0/16 means that the top (leftmost) 16 bits of theaddress 128.6.0.0 identify the subnetwork. The subnetworklogically divides an IP address into a network part (thebits that make up the subnet) and the host part (thebits that identify the host within the subnet).

(also called a netmask) is a bit mask that contains ones in thepositions of the network bits of the address. For Rutgers,this means the top 16 bits will be one, resulting in asubnet mask of 255.255.0.0. A subnet mask is used tostrip the host bits from the address to match prefixes ina forwarding table.

A subnet mask is used tostrip the host bits from the address to match prefixes ina forwarding table. For Rutgers,this means the top 16 bits will be one, resulting in asubnet mask of 255.255.0.0. For instance, the host aramis.rutgers.edu has an address of128.6.4.2 and a netmask of 0xffffff00. This indicates thatit is in a subnetwork that is defined by the prefix128.6.4.0/24.

An address of 255.255.255.255. This is a broadcast address for the host’s network.Datagrams directed to this address will be delivered to allhosts on the directly-connected network but routers will notforward them to other networks (they are limited to the same localarea network asthe sender).

.An address with only the host bits set to one (e.g., 128.6.255.255).Datagrams directed to this address will be routed to thespecified subnet (if the router permits it) and delivered to allhosts on that subnet (they are directed to a specific subnet).Routers may be configured to forwardthese datagrams to ensure that they are delivered to subnetsoutside the directly-connected local area network.

Its IP address, Its subnet mask., Its gateway, One or more domain name servers

so it can identify itself in the source address field of an IP header.

Using the subnet mask along with the IP address, it can identify its own subnet and hence identify which addresses are on the local subnet and which ones need to be directed to a router.

This is a router on the LAN and the defaultaddress for non-local addresses that are not in a host’s localrouting table. A gateway is a simple router that routes datagramsbetween the LAN and another network.

It needs to know the addressof at least one name server so that it can look up Internetdomain names and find corresponding addresses

These four parameters can be configured manually. Alternatively,the Dynamic Host Configuration Protocol (DHCP) canbe used to do this automatically.

Dynamic Host Configuration Protocol (DHCP). DHCP is a protocol to allow a client to get an IP address foritself as well as essential network configuration parameters.

developing such a protocol is that it hasto work before the client has a valid address on the network.Hence, a conventional request-response protocol with sourceand destination addresses will not work. A requirement forDHCP is that the DHCP server has to be running on the samelocal area network as the host. If not, a DHCP Relay Agent mustrun that serves as a proxy and forwards requests and responsesto the remote DHCP server.

the protocol hasto work before the client has a valid address on the network.

the DHCP server has to be running on the samelocal area network as the host. If not, a DHCP Relay Agent mustrun that servers as a proxy and forwards requests and responsesto the remote DHCP server.

a DHCP Relay Agent mustrun that servers as a proxy and forwards requests and responsesto the remote DHCP server.

DHCP uses limited broadcast messages (255.255.255.255). Aclient is allowed to send a limited broadcast and is capableof receiving one even if does not have an address assigned. They use UDP style messages.

DORA. Discover. offer. request. acknowledge.

Discover. The client sends a limited broadcast"DHCP Discover" UDP message to port 67. This contains arandom transaction identifier.

Offer. The server listens to broadcasts coming in onport 67. It gets the Discover message and responds backby sending a limited broadcast "DHCP Offer" UDP messageto port 68. The response contains the following parameters:Matching transaction, identifierProposed IP address,Subnet mask,Lease time

Request. The client picks up the server’s Offermessage. It compares the transaction identifier to ensurethat the offer is not directed to another client. If therehave been multiple DHCP servers and it received multipleoffers, it selects the one it wants to accept and ignoresthe others. The client responds with a Request messagethat contains a copy of the parameters in the Offer.

ACK. The server associates the offered parameterswith the host and sends back a DHCP ACK messageacknowledging the association. The client can now configureits network with those parameters.

Matching transaction identifier,Proposed IP address,Subnet mask,Lease time

Automatic allocation. dynamic allocation. manual allocation.

DHCP can be used to assigna permanent IP address to a host.

DHCP can be used to leasean address to a host. The host may use the address fora specified period of time. This allows the reuse of anaddress after it is no longer needed by the host.A Wi-Fi hotspot is a common example of this use of DHCP.

wifi hotspot

An administrator can configurethe DHCP server to assign a specific address in responseto a DHCP Discover message. This is done by associatingthe host’s link layer address (e.g., Ethernet MAC address)with a specific IP address.

the massive need for unique IP addresses.

by allowing an organization to create aprivate IP address space within the organizationwhile presenting one, or a small set of IP addressesto the outside Internet. As a packet flows througha NAT-enabled router, the router uses aNAT Translation Table to map a source{private-address, port1} to a {public-address, port2}.

a source{private-address, port1} to a {public-address, port2} and When packets flow back to the router from the outside,the router uses the NAT Translation Table to performthe inverse mapping of{public-address, port2} to {private-address, port1}.

the gateway router has to look at, and possiblymodify, the transport layer header since since a sourceport number may need to be changed to one that is not usedby any other internal-external mapping at the router.

private addresses are non-routable on the Internet and can onlybe used in internal networks. RFC 1918 defines threeaddress blocks that can be used for these addresses.

NAT is not particularly useful for serversbut is incredibly useful for client machines. Hosts in a NAT environment cannot accept incoming packetsunless a host/port mapping has been established by an outgoingpacket.

Internet Control Message Protocol (ICMP) is a simple network-layer protocol that was designed to allowhosts and routers to communicate network-related information

an eight byte or greater segment that sits in the payload(data section) of an IP datagram. designed to allowhosts and routers to communicate network-related information

It contains a checksum over the Internet Control Message Protocol (ICMP) header and associateddata as well astype and code fields,which define the purpose of the message.Depending on the message, four additional bytes may specifyparameters to the message and optional data may containthe IP header and first eight bytes of the original datagramfor which ICMP is generating a report.

defining the type of message. eg: an echo request (ping),echo response (ping), a destination unreachable status,a TTL exceeded warning, and a bad IP header error

example of a service that uses Internet Control Message Protocol (ICMP) .It creates a raw socket and generates an ICMP message of thetype echo request (type 8). When the message is routedto the destination host, the ICMP protocol sends back anICMP echo reply (type 0) datagram.

traces a route to a specific host. uses Internet Control Message Protocol (ICMP) by sending a series of UDP segments to abogus destination port on the desired host. Each UDP segmenthas a progressively longer time-to-live (TTL) value in theIP header. The first router will not route the datagram witha TTL of 1 since it decremented to 0 and hence expired.Instead, the router sends back an ICMP TTL exceededwarning message that contains the name and address of the router inthe body of the ICMP message. The datagram with a TTL=2 will be routedby the first router but will be rejected by the second one, and so on.

the Time to live

a checksum. The designers reasoned that the link layer has achecksum and TCP as well as UDP include critical IP fields intheir checksum computation.

fragmentation fields, because with ipv6 routers will never fragment datagrams.

systems with both network stacks implementedand capable of using either protocol. (ipv4 or ipv6) IPv6 systems can bridge toIPv4 systems since the IPv4 address space is mapped onto asubset of the IPv6 space. The problem is that IPv4 systemscannot effectively communicate with IPv6 systems due to itslarger address space. A system using IPv6 may not be visibleto a system on an IPv4 network.

IPv6 uses a huge address space: 128-bit addresses comparedwith IPv4’s 32-bit addresses. A 128-bit address allows for3.4×1038 addresses, which is 8.9×1028times more than IPv4. Even though its addresses are longer,IPv6 uses a simplified header compared to its predecessor.It is a fixed-length headers with fewer fields.