Page 3 of 11

A BRIEF INTRODUCTION TO INTERNET ADDRESSING AND ROUTING

Internet protocol addressing and routing must first be understood before any discussion of IP address space design issues will be useful. This section goes over the original IP address scheme, its limitations, and the current methods used to deal with the finite number of IP addresses. This information is crucial to an understanding of the choices and constraints for IP addresses in a data center.

Table 1: Original IP version 4 address class

Table 1 shows the different classes of IP addresses. Note that two other classes of address space, class D and E, were not included in the above table. Class D addresses start at 224.0.0.0 and are used for multicast. Class E addresses start at 240.0.0.0 and are used for experimental purposes.

Original IP 4 Addressing Scheme
In order for two hosts to communicate over the Internet, there needs to be a way to uniquely identify hosts. In 1981, the Internet Engineering Task Force (IETF) created the Internet Protocol, IP version 4 (IPv4) [1], which defines the current method of uniquely identifying hosts. IPv4 addressing uses a 32-bit binary address. The IETF also incorporated support for decimal representation of addresses to make the addresses human-readable. In decimal form, an IP address consists of 4 octets (sets of 8 bits), separated by dots. Each octet can be a number ranging from 0 to 255. Examples of valid (decimal) IP addresses are 10.245.171.1 and 172.16.50.224.

IP addresses are partitioned into a network portion followed by a host portion. Hosts belong to a network, and that network is defined by the network portion of the IP address. The original design called for classes of address space that divided the IP space into large, medium, and small networks that could be assigned to organizations (businesses, universities, government agencies, etc). Included in the design was the notion of a network mask that defines what part of an IP address is the network portion (as opposed to the host portion of the address). In binary, the network portion of the address is a series of ones that is then followed by a series of zeroes representing the host portion of the address. In decimal, the network portion of the mask is equal to 255 for each octet.

Autonomous Systems
Another requirement for Internet communication is that each host needs to know how to reach all of the other hosts. To facilitate this, organizations advertise the path to their network to other networks. Devices called routers learn about networks in this way and forward packets appropriately. Routers exchange network and routing information through what is called a routing protocol. A routing table, which is a list of networks and the next hop (often another router) to forward packets for those networks, is stored in the router’s memory. The router will select the best path (next-hop router) to put into its table when multiple paths exist to the same network.

In the early days of the Internet, all connected routers shared their routing tables. As use of the Internet started to grow, more routers and networks were added, and the amount of overhead required to store the routing table and manage changes to the routing table also increased. In addition, as more companies began manufacturing different routers that ran their own implementation of the routing software, compatibility issues between different vendors arose. For these and a number of other reasons, it was decided to break the Internet into smaller routing domains, called Autonomous Systems (AS).

An autonomous system (AS) [2] is a set of routers and networks that are managed by one or more administrative entities (e.g., company, university, Internet Service Provider, etc.). Each AS is assigned a unique number so that communication between different autonomous systems can occur. Routers inside the AS run an Interior Gateway Protocol (IGP) such as RIP [3] and OSPF [4]. To communicate externally, one or more border routers are chosen. Border routers use an Exterior Gateway Protocol (EGP) to exchange routing information with routers in different autonomous systems. Today, the Border Gateway Protocol Version 4 [5] (BGP4) is generally used for this purpose.

Each AS has a number associated with it. BGP4 uses 16 bits for AS numbers, so that AS numbers range from 0 to 64535. The upper 1024 are reserved as private AS numbers, usable only within an AS and not directly reachable from the Internet. This leaves AS numbers 1 to 64511 as valid, Internet-usable AS numbers.

Issues With IP Addressing
Since IPv4 was finalized, use of the Internet has grown exponentially, causing major addressing issues. In the early days of the Internet, organizations were able to obtain large blocks of IP space without proof that it was needed or even going to be used, and as a result, IP address space was being rapidly depleted. Another side effect of address space allocation policies was that the routing tables for Internet routers were once again becoming huge [6]. Remember, routers store a list of networks and next hop information in memory. When routing tables are large, they take up more memory and more CPU processing time is required to search them.

Finally, the class of address space as defined in Table 1 did not always meet, and sometimes exceeded, the needs of the organization receiving it. For example, a small business that expected to grow to no larger than 300 hosts would require two Class C networks (508 addresses). This wasted 208 addresses (two 256 host networks minus four addresses that are network overhead and minus the 300 hosts)!

Address Allocation Authority
To slow the depletion of IP space, the Internet Assigned Numbers Authority (IANA) [7] was established to oversee allocation of the remaining IPv4 addresses. IANA further delegated this authority to the regional registries:

• American Registry for Internet Numbers (ARIN)
• Asia-Pacific Network Information Center (APNIC)
• Réseau IP Européens (RIPE NCC)

Subnetting Changes
Several new methods of addressing were also created so that usage of IP space was more efficient. The first of these methods is called Variable-Length Subnet Masking (VLSM) [8]. Subnetting had long been a way to better utilize address space [9]. Subnets divide a single network into smaller pieces. This is done by taking bits from the host portion of the address to use in the creation of a “sub” network. For example, take the class B network 147.208.0.0. The default network mask is 255.255.0.0, and the last two octets contain the host portion of the address. To use this address space more efficiently, we could take all eight bits of the third octet for the subnet.

One drawback of subnetting is that once the subnet mask has been chosen, the number of hosts on each subnet is fixed. This makes it hard for network administrators to assign IP space based on the actual number of hosts needed. For example, assume that a company has been assigned 147.208.0.0 and has decided to subnet this by using eight bits from the host portion of the address. Assume that the address allocation policy is to assign one subnet per department in an organization. This means that 254 addresses are assigned to each department. Now, if one department only has 20 servers, then 234 addresses are wasted.

Using variable-length subnet masks (VLSM) improves on subnet masking. VLSM is similar to traditional fixed-length subnet masking in that it also allows a network to be subdivided into smaller pieces. The major difference between the two is that VLSM allows different subnets to have subnet masks of different lengths. For the example above, a department with 20 servers can be allocated a subnet mask of 27 bits. This allows the subnet to have up to 30 usable hosts on it.

Table 2: Private address space ranges

Private IP Space
In 1996, IANA set aside three blocks of the global IP space to be used by organizations solely for the purpose of internal communication [10]. This address space, called private IP space, meant that a company could assign private addresses to hosts inside the company that did not require direct access to the Internet. Any organization could use private space without fear of colliding with another organization’s address space. This allowed companies to conserve on the public IP space they had already acquired by assigning it to only those hosts that needed to communicate directly with the Internet. Table 2 shows which networks can be used for private addressing.

Classless Internet Domain Routing
So far, the discussion on IP address allocation has used the model shown in Table 1. This model is often referred to as a “classful” model because it relies on using the definitions of class A, B, and C networks. Classless Inter-Domain Routing (CIDR) [11, 12] eliminates classful addressing in the same way that VLSM eliminated fixed-length subnet masks. CIDR uses a prefix to indicate the number of bits used for the network portion of the address, while the remaining bits are used for the host address. For example, 147.208.61.8/20 is a CIDR address in which the first 20 bits contain the network portion of the address, leaving 12 bits for the host portion. The network mask for a /20 prefix is 255.255.240.0 and is equivalent to 16 traditional class C networks!

Another advantage of CIDR is it allows routes to be aggregated. This means many networks can be summarized into a single route. For example, 147.208.0.0/19, 147.208.32.0/19, 147.208.64.0/19, and 147.208.192.0/19 can be summarized as 147.208.0.0/17. Once CIDR was implemented, the growth in the size of Internet routing tables was significantly reduced.