Subnetting •32 bit addresses carefully assigned to make the IP address of all hosts on a given network share a common prefix. •Advantage: Reduces size of routing tables within the Internet •Problems with the basic 2 level addressing mechanism: •Naming: there’s not a one-to-one correspondence between a host and an address/route. •Flexibility and scalability: Each physical network requires a unique network address. 1 Subnetting •Naming problems: •The path taken by a multihomed host depends on the address specified. 2 Subnetting •Flexibility and scalability example: Consider a medium Enterprise (8000 hosts) that wants >1 physical networks. •Options: purchase multiple class B addresses, one class B plus additional blocks of Class C address, or limit to one physical network. 3 Subnetting •Other option is to extend the basic two level addressing mechanism. •Any extension must abide by the following: •All hosts and routers at the site must agree to honor the sites addressing scheme •Other sites on the Internet treat the addresses as in the original scheme. Huge problem: shortage of Class B addresses. Objective: Minimize the number of Class B network addresses without destroying the original scheme. 4 Subnetting Subnet addressing: Allows an organization to create multiple logical networks internally while preserving the presence of a single autonomous network to the outside. An autonomous network: A collection of routers and networks that fall under one administrative entity. 5 Subnetting Subnetting creates a 3 level addressing hierarchy •Subnet mask identifies the extended network prefix. •Subnetting is defined by RFC 950 and is a required Internet standard. 6 Subnetting Example : 130.127.49.225 mask 255.255.255.0 10000010 01111111 00110001 11100001 Network ID Network Number Network Prefix Host ID Extended Network Prefix Subnetwork Address Subnet Number SubnetID Classfull network address: 130.127.0.0 Classfull network prefix notation: 130.127/16 Subnetwork address: 130.127.49.0 Network Prefix Notation: 130.127.49/24 7 Subnetting Example : Subnetwork design 130.127/24 8 Subnetting Example using bitwise AND operation to learn the extended network prefix (using host IP address: 130.127.49.225, mask 255.255.255. ) 10000010 01111111 00110001 11100001 11111111 11111111 11111111 00000000 -------------------------------------------------------------------10000010 01111111 00110001 00000000 Subnet id: 49, host id: 225 Subnetwork address in network prefix notation: 130.127.49/24, Classfull Internet Network Address: 130.127/16 Subnet broadcast on this subnet would be: •130.127.49.255 •<130.127, 49, 255> 9 Subnetting Network Design Example: 3 networks, 2000, 6000 and 1000 hosts. Use a single Class B address. But expect the number of networks to grow to 5 in the future. Design the IP addressing scheme. 10 Subnetting Additional notation: {<network number>, <subnet number>, <host number> } Example: write the address 130.127.49.205/19 in this notation: Although this notation does not indicate the number of subnets, it is useful for broadcasts. What is the broadcast for this subnet ? 11 Subnetting •Remember the all 1’s and all 0’s… we will assume that the number of valid subnets and hosts (based on subnet bits) does not include the all 1’s and all 0’s. •To allow maximum flexibility in choosing how to partition subnet addresses, RFC 950 permits subnet interpretation to be chosen independently for each physical network. All machines must honor it! 12 Address Resolution The address resolution problem: remember IEEE 802.2 encapsulation... Dest Address (6) Source Address (6) Length (2) DSAP SSAP cntl code type (1) (1) (1) (3) (2) Data ( 38 - 1492) CRC (4) 13 Address Resolution Resolution through direct binding: Phys = f(IP) • Objective is to encode the IP address in Physical Dynamic binding protocol is required for Ethernet networks where: •Physical address larger than IP address (48 vs 32 bits) •Physical address can change •Potentially many hosts on the network Address Resolution Protocol (ARP) defined by RFC 826 14 Address Resolution •ARP is a protocol outside the set of TCP/IP protocols 15 Address Resolution ARP Protocol: Case 1: Host A sends to Host D 16 Address Resolution Case 1 Issues •Why might there not be a response from D ? •What happens if D does not respond? •ARP cache entries timeout after 20 minutes •Why? •Why is the entry removed regardless of if the host uses the information during the 20 minutes? 17 Address Resolution Case 2: Host D receives an ARP request message from Host A with target Host D. Case 3: Host C observes the request from A to D. It will use it only if A’s mapping is in its cache AND it is different. (See stevens TCP Illustrated vol 1) ARP topics: •Arp command •Gratuitous arp •Arp security issues 18 Address Resolution Reverse Address Resolution Protocol •Used when a diskless host boots to learn its IP address •There are other ways a Host can learn its Address: Bootp and DHCP 19