Network Layer We have discussed data link architectures CSMA/CD Point-to-Point Wireless LANs These architectures deliver frames to next station They implement network segments, possibly connected through bridges One might build a private medium size private network this way Could not build a universal or public network this way
Network Layer We would like to interconnect such networks into a consistent larger network We could consider this larger network an internet, an interconnection of networks Logical ‘gluing’ of heterogeneous layer 2 networks to appear as a single network A number of issues must be addressed Universal addressing Routing Packetizing data
Network Layer Physical Layer Twisted pairs Fiber Optics encoding CSMA/CD IEEE802.3 Pt-to-Pt HDLC PPP IEEE Network Layer Addressing Routing Packetizing
Network Layer There are and have been many network layer protocols Systems Network Architecture (SNA) – IBM DECNET – Digital Equipment Corporation OSI – International Standards Organization AppleTalk – Apple Computer Internetwork Packet Exchange (IPX) – Novell Internet Protocol (IP) Each of these has its own addressing scheme Each has own packet format Each implements own versions of routing
Network Layer We will discuss the Internet Protocol Recall, this is layered protocol, much like the OSI model Internet Protocol does not have its own physical or data link layers Uses existing lower layer protocols as discussed The Internet Protocol is often called TCP/IP which represents its two underlying protocols
Comparison of OSI and TCP/IP Data Link Physical
Internet Protocol Historical Summary 1969 – Four node ARPANET established 1973 – Development of TCP/IP suite begins 1978 – UNIX distributed to academic sites 1981 – CSNET established 1983 – TCP/IP becomes official protocol 1983 – NSF funds a national backbone linking 6 supercomputer centers Emerging regional networks link to backbone nodes Initial backbone 56Kbps
Internet Protocol Historical Summary 1987 – NSF backbone increased to T1 Partnership formed to operate NSFnet Merit – Michigan IBM MCI 1991 NSFNET backbone increased to 45 Mbps 1993 – NSFNET establishes migration to a commercial Internet 1995 – Internet Service Providers (ISPs) established
Internet 2 Established in 1998 Separate physical network for research institutions Uses same Internet Protocol Network Operations Center (noc) at Indiana University Core Network called Abilene Circuits provided by Qwest Communications
Abilene Access Nodes Atlanta Indianapolis Kansas City Denver Los Angeles Sacramento Seattle Abilene Core Node Abilene Access Node Operational January 1999 Houston Cleveland New York
Internet today
Internet Standards Bodies Internet Society (ISOC ) Internet Architecture Board (IAB ) Internet Engineering Task Force (IETF) Internet Research Task Force (IRTF) Internet Protocols Routing Management
Internet Standards Documentation Official standards published in documents called Request for Comments (RFCs) RFCs go through various stages Internet Draft Experimental Informational Proposed standard –Draft Standard –Internet Standard RFCs are stored in a public repository
Internet Addressing For an Internet, we need universal identification of nodes Must apply to different types of networks Must be independent of vendor, hardware, physical network In TCP/IP we deal with various addresses Physical addresses – MAC addresses IP addresses – universal address Port – specifies service
Internet Addressing Administration of Internet addresses Internet Corporation for Assigned Names and Numbers (ICANN) Internet Assigned Numbers Authority (IANA) Authorizes and oversees 5 registries APNIC (Asia Pacific Network Information Centre) ARIN (American Registry for Internet Numbers) RIPE NCC (Réseaux IP Européens) LACNIC (Regional Latin-American and Caribbean IP Address Registry)LACNIC (Regional Latin-American and Caribbean IP Address Registry) AfriNIC (African Network Information Centre)
Internet Address An IP address is 32 bit word Maximum of 2 32 or 4,294,967,296 addresses Address are assigned in ranges to accommodate Large networks Medium size networks Small networks Convenient to break address into two parts Range number – network Individual numbers within the range - host
Internet Address Address space originally defined into classes Class A - Maximum 128 Class A networks - Each network could have 2 24 or 16,777,215 addresses or hosts Class B - Maximum 16,384 Class B networks - Each has 65,536 addresses or nodes 0netidhostid netid hostid Net id = Net id =
Internet Address Class C - Maximum 2,097,152 Class C networks - Each network has 2 8 or 256 addresses Class D Class E Net id = Net id = netid hostid multicast future 310
Netid and hostid
Internet Addresses It is common place to represent a 32 bit Internet address in dotted notation Given a 32 bit address 0x826F27A2 We would write as Here is the network address is the host id The address in this network range to Likewise, the is part of the range to
Special Internet Addresses By convention 0 means ‘this’ 1 means ‘all’ All 0’s 0’s hostid 127 Anything This host Host on this network Loopback Address Data sent to this address is returned Never sent out on network
Special Internet Addresses netid 0s Refers to this network (netid) Not assigned to any host netid 1s All hosts on this network (netid) Called a directed broadcast 0s 1s All host on local network Called a limited broadcast
Addresses and Routing Devices that determine paths or routing need only know about network addresses Only the destination network need be concerned about the host address Devices that determine paths or routes are usually called routers Routers must have tables entries, called a routing table, for every network in order to determine paths
Subnetting Some networks (Class A & B) are large enough to warrant breakdown into smaller groups, subnets An organization may be assigned a large range (Class B) and may wish to allocate to departments in smaller subnets To make internal routers treat these subnets internally as separate networks requires additional information Must specify which bits in the address represents the network id
Subnetting To provide for subnetting addresses are specified in two parts The address A mask – indicate which bits make up network address Example: is part of a class B We could break this into smaller networks would be treated as a network id and.150 as the host id within that network
Subnetting Specified in RFC 950 Examples
Subnetting Alternative notation – Slash notation Instead of specifying a mask, indicate how many bits constitute the network address /24 Thus, any Class B network could be specified as a.b.c.d/16
Supernetting Much like subnetting a Class x network, we can collect adjacent Class x networks to form larger networks Consider the two Class C networks Each of these is a network of 256 addresses We could group these into a single network as follows /23
Determining Network/Host Address AND AND
Defining Subnets An organization is allocated a Class B network The organization needs at least 1000 subnets If a common mask will be used for all these subnets, the number must be a power of 2. There will be 1024 subnets The mask will be Each subnet will have 64 addresses
Defining Subnets Host address (64) Subnet (1024) Network ID
Defining Subnets What is the range of the first subnet? What is the range of the second subnet? What is the range of the last subnet? What is the range of the 1023 rd subnet?
Classfull Addressing Class A networks are too large which results into wasted address space Class B are also too large for many organizations Class C networks are too small, requiring multiple allocations to the same organization In the early 90s, it became clear that this method of address space allocation would lead to early depletion of addresses
Classless Interdomain Routing (CIDR) To preserve address space, a new method of assigning space was developed in 1993 Groups of unassigned address were allocated regionally (RFC 1466) New allocation scheme provided for allocation of variable length blocks of addresses Going forward, allocations would no longer be done by Class Allocation scheme defined in RFC 1518
Classless Interdomain Routing Address blocks are sized by powers of 2 Blocks of size 2,4,8,…., 256, 512, 1024… can be assigned Stating address must be divisible by the number of addresses in the block The allocation is specified the same fashion as subnets using slash notation Original allocations are maintained and folded into this scheme Routers modified to accept and maintain their routing tables in this way
Managing Address Space Blocks of addresses assigned to ISPs ISPs assign subnets to customers Assigned subnets can be of variable sizes When customers change Service Provider, they usually have to change address space Usually, the ISP will assign large subnets first, followed by smaller subnets
Variable size Subnetting An ISP is granted a block of addresses starting with /16. The ISP needs to distribute these addresses to three groups of customers as follows: Group 1 has 64 customers; each needs 256 addresses Group 2 has 128 customers; each needs 128 addresses. Group3 has 128 customers; each needs 64 addresses.
Variable size Subnetting Group 1 For this group, each customer needs 256 addresses. This means the suffix (host-id) length is 8 The prefix length (net-id) is then = : /24 /24 02: /24 /24 ………………………………….. 64: /24 /24 Total = 64 256 = 16,384
Variable size Subnetting Group 2 For this group, each customer needs 128 addresses. This means the suffix length is 7 (27 = 128). The prefix length is then = 25. The addresses are: 001: /25 /25 002: /25 /25 ………………………………………. 128: /25 /25 Total = 128 128 = 16,384
Variable size Subnetting Group 3 For this group, each customer needs 64 addresses. This means the suffix length is 6 (26 = 64). The prefix length is then = : /26 /26 002: /26 /26 ………………………… 128: /26 /26 Total = 128 64 = 8,192
UMS Managed Address Space /16University of Maine System /16Maine School and Library Network /16State of Maine Government /24Jackson Laboratories /24Waterville K /24State of Maine Government /23State of Maine Government /24College of the Atlantic /19University of Maine System /19University of Maine System /18University of Maine System /18University of Maine System /19University of Maine System /20University of Maine System
Private Addresses Some organizations want to establish IP networks internally, but not outside the organization By convention, some address space is allocated for this purpose These addresses are not accepted (routed) on the Internet, but can be routed internally
Private Addresses RFC 1918 defines this address space to to to
IP addressing tools Web tool for calculating subnet masks Tool for calculating subnet masks ipcalc321.exe on course web site(Wildpackets) Tool to look up ownership of IP address space cyberkit - cyber30.zip on web site