Before you can send a message, you must know the destination address. It is extremely important to understand that each computer has several addresses, each used by a different layer. One address is used by the data link layer, another by the network layer, and still another by the application layer.
When users work with application software, they typically use the application layer address. For example, in next topic, we discussed application software that used Internet addresses (e.g., www.indiana.edu). This is an application layer address (or a server name). When a user types an Internet address into a Web browser, the request is passed to the network layer as part of an application layer packet formatted using the HTTP protocol (Figure 5.6).
The network layer software, in turn, uses a network layer address. The network layer protocol used on the Internet is IP, so this Web address (www.indiana.edu) is translated into an IP address that is 4 bytes long when using IPv4 (e.g., 129.79.127.4) (Figure 5.6). This process is similar to using a phone book to go from someone’s name to his or her phone number.2
The network layer then determines the best route through the network to the final destination. On the basis of this routing, the network layer identifies the data link layer address of the next computer to which the message should be sent. If the data link layer is running Ethernet, then the network layer IP address would be translated into an Ethernet address. Next topic shows that Ethernet addresses are six bytes in length, so a possible address might be 00-0F-00-81-14-00 (Ethernet addresses are usually expressed in hexadecimal) (Figure 5.6).
|
Address |
Example Software |
Example Address |
|
Application layer |
Web browser |
www.kelley.indiana.edu |
|
Network layer |
Internet Protocol |
129.79.127.4 |
|
Data link layer |
Ethernet |
00-0C-00-F5-03-5A |
Figure 5.6 Types of addresses
Assigning Addresses
In general, the data link layer address is permanently encoded in each network card, which is why the data link layer address is also commonly called the physical address or the MAC address. This address is part of the hardware (e.g., Ethernet card) and can never be changed. Hardware manufacturers have an agreement that assigns each manufacturer a unique set of permitted addresses, so even if you buy hardware from different companies, they will never have the same address. Whenever you install a network card into a computer, it immediately has its own data link layer address that uniquely identifies it from every other computer in the world.
Network layer addresses are generally assigned by software. Every network layer software package usually has a configuration file that specifies the network layer address for that computer. Network managers can assign any network layer addresses they want. It is important to ensure that every computer on the same network has a unique network layer address so every network has a standards group that defines what network layer addresses can be used by each organization.
Application layer addresses (or server names) are also assigned by a software configuration file. Virtually all servers have an application layer address, but most client computers do not. This is because it is important for users to easily access servers and the information they contain, but there is usually little need for someone to access someone else’s client computer. As with network layer addresses, network managers can assign any application layer address they want, but a network standards group must approve application layer addresses to ensure that no two computers have the same application layer address. Network layer addresses and application layer addresses go hand in hand, so the same standards group usually assigns both (e.g., www.indiana.edu at the application layer means 129.79.78.4 at the network layer). It is possible to have several application layer addresses for the same computer. For example, one of the Web servers in the Kelley School of Business at Indiana University is called both www.kelley.indiana.edu and www.kelley.iu.edu.
Internet Addresses No one is permitted to operate a computer on the Internet unless they use approved addresses. ICANN (Internet Corporation for Assigned Names and Numbers) is responsible for managing the assignment of network layer addresses (i.e., IP addresses) and application layer addresses (e.g., www.indiana.edu). ICANN sets the rules by which new domain names (e.g., com, .org, .ca, .uk) are created and IP address numbers are assigned to users. ICANN also directly manages a set of Internet domains (e.g., .com, .org, .net) and authorizes private companies to become domain name registrars for those domains. Once authorized, a registrar can approve requests for application layer addresses and assign IP numbers for those requests. This means that individuals and organizations wishing to register an Internet name can use any authorized registrar for the domain they choose, and different registrars are permitted to charge different fees for their registration services. Many registrars are authorized to issue names and addresses in the ICANN managed domains, as well as domains in other countries (e.g., .ca, .uk, .au).
Several application layer addresses and network layer addresses can be assigned at the same time. IP addresses are often assigned in groups, so that one organization receives a set of numerically similar addresses for use on its computers. For example, Indiana University has been assigned the set of application layer addresses that end in indiana.edu and iu.edu and the set of IP addresses in the 129.79.x.x range (i.e., all IP addresses that start with the numbers 129.79).
In the old days of the Internet, addresses used to be assigned by class. A class A address was one for which the organization received a fixed first byte and could allocate the remaining three bytes. For example, Hewlett-Packard (HP) was assigned the 15.x.x.x address range which has about 16 million addresses. A class B address has the first two bytes fixed, and the organization can assign the remaining two bytes. Indiana University has a class B address, which provides about 65,000 addresses. A class C address has the first three bytes fixed with the organization able to assign the last byte, which provides about 250 addresses.
People still talk about Internet address classes, but addresses are no longer assigned in this way and most network vendors are no longer using the terminology. The newer terminology is classless addressing in which a slash is used to indicate the address range (it’s also called slash notation). For example 128.192.1.0/24 means the first 24 bits (three bytes) are fixed, and the organization can allocate the last byte (eight bits).
One of the problems with the current address system is that the Internet is quickly running out of addresses. Although the four-byte address of IPv4 provides more than 4 billion possible addresses, the fact that they are assigned in sets significantly limits the number of usable addresses. For example, the address range owned by Indiana University includes about 65,000 addresses, but we will probably not use all of them.
The IP address shortage was one of the reasons behind the development of IPv6, discussed previously. Once IPv6 is in wide use, the current Internet address system will be replaced by a totally new system based on 16-byte addresses. Most experts expect that all the current four-byte addresses will simply be assigned an arbitrary 12-byte prefix (e.g., all zeros) so that the holders of the current addresses can continue to use them.
Subnets Each organization must assign the IP addresses it has received to specific computers on its networks. In general, IP addresses are assigned so that all computers on the same LAN have similar addresses. For example, suppose an organization has just received a set of addresses starting with 128.192.x.x. It is customary to assign all the computers in the same LAN numbers that start with the same first three digits, so the business school LAN might be assigned 128.192.56.x, which means all the computers in that LAN would have IP numbers starting with those numbers (e.g., 128.192.56.4, 128.192.56.5, and so on) (Figure 5.7). The computer science LAN might be assigned 128.192.55.x, and likewise, all the other LANs at the university and the BN that connects them would have a different set of numbers. Each of these LANs is called a TCP/IP subnet because computers in the LAN are logically grouped together by IP number.
Routers connect two or more subnets so they have a separate address on each subnet. The routers in Figure 5.7, for example, have two addresses each because they connect two subnets and must have one address in each subnet.
Although it is customary to use the first three bytes of the IP address to indicate different subnets, it is not required. Any portion of the IP address can be designated as a subnet by using a subnet mask.
Figure 5.7 Address subnets
Every computer in a TCP/IP network is given a subnet mask to enable it to determine which computers are on the same subnet (i.e., LAN) that it is on and which computers are outside of its subnet. Knowing whether a computer is on your subnet is very important for message routing, as we shall see later in this topic.
For example, a network could be configured so that the first two bytes indicated a subnet (e.g., 128.184.x.x), so all computers would be given a subnet mask giving the first two bytes as the subnet indicator. This would mean that a computer with an IP address of 128.184.22.33 would be on the same subnet as 128.184.78.90.
IP addresses are binary numbers, so partial bytes can also be used as subnets. For example, we could create a subnet that has IP addresses between 128.184.55.1 and 128.184.55.127, and another subnet with addresses between 128.184.55.128 and 128.184.55.254.
Dynamic Addressing To this point, we have said that every computer knows its network layer address from a configuration file that is installed when the computer is first attached to the network. However, this leads to a major network management problem. Any time a computer is moved or its network is assigned a new address, the software on each individual computer must be updated. This is not difficult, but it is very time consuming because someone must go from office to office editing files on each individual computer.
The easiest way around this is dynamic addressing. With this approach, a server is designated to supply a network layer address to a computer each time the computer connects to the network. This is commonly done for client computers but usually not done for servers.
The most common standard for dynamic addressing is Dynamic Host Configuration Protocol (DHCP). DHCP does not provide a network layer address in a configuration file. Instead, there is a special software package installed on the client that instructs it to contact a DHCP server to obtain an address. In this case, when the computer is turned on and connects to the network, it first issues a broadcast DHCP message that is directed to any DHCP server that can "hear" the message. This message asks the server to assign the requesting computer a unique network layer address. The server runs a corresponding DHCP software package that responds to these requests and sends a message back to the client giving it its network layer address (and its subnet mask).
The DHCP server can be configured to assign the same network layer address to the computer (on the basis of its data link layer address) each time it requests an address, or it can lease the address to the computer by picking the "next available" network layer address from a list of authorized addresses. Addresses can be leased for as long as the computer is connected to the network or for a specified time limit (e.g., 2 hours). When the lease expires, the client computer must contact the DHCP server to get a new address. Address leasing is commonly used by ISPs for dial-up users. ISPs have many more authorized users than they have authorized network layer addresses because not all users can log in at the same time. When a user logs in, his or her computer is assigned a temporary TCP/IP address that is reassigned to the next user when the first user hangs up.
Dynamic addressing greatly simplifies network management in non-dial-up networks, too. With dynamic addressing, address changes need to be made only to the
Subnet Masks
TECHNICAL FOCUS
Subnet masks tell computers what part of an Internet Protocol (IP) address is to be used to determine whether a destination is on the same subnet or on a different subnet. A subnet mask is a four-byte binary number that has the same format as an IP address. A 1 in the subnet mask indicates that that position is used to indicate the subnet. A 0 indicates that it is not.
A subnet mask of 255.255.255.0 means that the first three bytes indicate the subnet; all computers with the same first three bytes in their IP addresses are on the same subnet. This is because 255 expressed in binary is 11111111.
In contrast, a subnet mask of 255.255.0.0 indicates that the first two bytes refer to the same subnet.
Things get more complicated when we use partial-byte subnet masks. For example, suppose the subnet mask was 255.255.255.128. In binary numbers, this is expressed as:
This means that the first three bytes plus the first bit in the fourth byte indicate the subnet address.
Likewise, a subnet mask of 255.255.254.0 would indicate the first two bytes plus the first seven bits of third byte indicate the subnet address, because in binary numbers, this is:
The bits that are ones are called network bits because they indicate which part of an address is the network or subnet part, while the bits that are zeros are called host bits because they indicate which part is unique to a specific computer or host.
DHCP server, not to each individual computer. The next time each computer connects to the network or whenever the address lease expires, the computer automatically gets the new address.
Address Resolution
To send a message, the sender must be able to translate the application layer address (or server name) of the destination into a network layer address and in turn translate that into a data link layer address. This process is called address resolution. There are many different approaches to address resolution that range from completely decentralized (each computer is responsible for knowing all addresses) to completely centralized (there is one computer that knows all addresses). TCP/IP uses two different approaches, one for resolving application layer addresses into IP addresses and a different one for resolving IP addresses into data link layer addresses.
Server Name Resolution Server name resolution is the translation of application layer addresses into network layer addresses (e.g., translating an Internet address such as www.yahoo.com into an IP address such as 204.71.200.74). This is done using the Domain Name Service (DNS). Throughout the Internet a series of computers called name servers provides DNS services. These name servers have address databases that store thousands of Internet addresses and their corresponding IP addresses. These name servers are, in effect, the "directory assistance" computers for the Internet. Anytime a computer does not know the IP number for a computer, it sends a message to the name server requesting the IP number. There are about a dozen high-level name servers that provide IP addresses for most of the Internet, with thousands of others that provide IP addresses for specific domains.
Whenever you register an Internet application layer address, you must inform the registrar of the IP address of the name server that will provide DNS information for all addresses in that name range. For example, because Indiana University owns the. indiana.edu name, it can create any name it wants that ends in that suffix (e.g., www.indiana.edu, www.kelley.indiana.edu, abc.indiana.edu). When it registers its name, it must also provide the IP address of the DNS server that it will use to provide the IP addresses for all the computers within this domain name range (i.e., everything ending in. indiana.edu). Every organization that has many servers also has its own DNS server, but smaller organizations that have only one or two servers often use a DNS server provided by their ISP. DNS servers are maintained by network managers, who update their address information as the network changes. DNS servers can also exchange information about new and changed addresses among themselves, a process called replication.
When a computer needs to translate an application layer address into an IP address, it sends a special DNS request packet to its DNS server.3 This packet asks the DNS server to send to the requesting computer the IP address that matches the Internet application layer address provided. If the DNS server has a matching name in its database, it sends back a special DNS response packet with the correct IP address. If that DNS server does not have that Internet address in its database, it will issue the same request to another DNS server elsewhere on the Internet.4
For example, if someone at the University of Toronto asked for a Web page on our server (www.kelley.indiana.edu) at Indiana University, the software on the Toronto client computer would issue a DNS request to the University of Toronto DNS server (Figure 5.8). This DNS server probably would not know the IP address of our server, so it would forward the request to the DNS root server that it knows stores addresses for the .edu domain.
Figure 5.8 How the DNS system works
The .edu root server probably would not know our server’s IP address either, but it would know that the DNS server on our campus could supply the address. So it would forward the request to the Indiana University DNS server, which would reply to the .edu server with a DNS response containing the requested IP address. The .edu server in turn would send that response to the DNS server at the University of Toronto, which in turn would send it to the computer that requested the address.
This is why it sometimes takes a longer to access certain sites. Most DNS servers know only the names and IP addresses for the computers in their part of the network. Some store frequently used addresses (e.g., www.yahoo.com). If you try to access a computer that is far away, it may take a while before your computer receives a response from a DNS server that knows the IP address.
Once your application layer software receives an IP address, it is stored on your computer in a DNS cache. This way, if you ever need to access the same computer again, your computer does not need to contact a DNS server. The DNS cache is routinely deleted whenever you turn off your computer.
Data Link Layer Address Resolution To actually send a message, the network layer software must know the data link layer address of the receiving computer. The final destination may be far away (e.g., sending from Toronto to Indiana). In this case, the network layer would route the message by selecting a path through the network that would ultimately lead to the destination. (Routing is discussed in the next section.) The first step on this route would be to send the message to its router.
To send a message to another computer in its subnet, a computer must know the correct data link layer address. In this case, the TCP/IP software sends a broadcast message to all computers in its subnet. A broadcast message, as the name suggests, is received and processed by all computers in the same LAN (which is usually designed to match the IP subnet). The message is a specially formatted request using Address Resolution Protocol (ARP) that says, "Whoever is IP address xxx.xxx.xxx.xxx, please send me your data link layer address." The software in the computer with that IP address then sends an ARP response with its data link layer address. The sender transmits its message using that data link layer address. The sender also stores the data link layer address in its address table for future use.5
