Routing TCP IP Volume II CCIE Professional Development

The Internet has been a huge success, driving the success of corporate internetworks. Few businesses are without Web sites these days (URLs can even be found on the corks of wine bottles), and e-mail is as important a business tool as the telephone. But certain aspects of IPv4 place an upper limit on how large the Internet can grow. 32 bits of address space limits the number of globally routable hosts that can connect and also limits the amount of hierarchy that can be created. As you have observed throughout much of this book, scalable internetworks require hierarchical routing. Hierarchical routing must be strictly maintained to enable the network to scale beyond the uses that application developers and Internet users are dreaming of today. To maintain hierarchy, Internet-connected sites must adhere to addressing and aggregation rules. Sites connected to an ISP or exchange usually must use addresses allocated to that ISP or exchange and reallocated to the site. This means that renumbering, with all the inherent difficulties described early in Chapter 2, "Introduction to Border Gateway Protocol 4," will remain an issue.

The success of the Internet also may increase data integrity, authenticity, and confidentiality requirements.

IPv4 network designers have alleviated some of these issues using a number of different techniques. As discussed in Chapter 4, "Network Address Translation," a network may use private addresses internally, using network address translation to communicate with theInternet or other companies, thereby mitigating the address space problem, allowing a huge number of nodes to access external internetworks. However, NAT is not always easy to implement and maintain. Some applications create excessive processing requirements on the NAT device, and other applications do not work at all. Furthermore, future Internet appliances, such as personal digital assistants, home security systems, or car maintenance computers, might require globally routable addresses so that they can be accessed from any Internet location.

The severe IPv4 hierarchy problems imposed by classful addresses were mitigated with the implementation of CIDR, as discussed in Chapter 2. CIDR enables you to group and divide more efficiently, but the total hierarchy is still limited to 32 bits of addressing space.

IPv6 addresses are so much bigger that there is enough address space for a large increase in globally routable addresses and for more layers of hierarchy. The size of the address space increased to 128 bits. Hierarchy is designed into the format of globally routable addresses.

ISPs assign a range of addresses to their clients. If the client wants to change ISPs, it most likely has to re-address its network. IPv4 network designers have implemented Dynamic Host Configuration Protocol (DHCP) to ease the burden of re-addressing PCs. DHCP works and will likely continue to be used with IPv6. IPv6 hosts can use DHCP or the built-in autoconfiguration method to configure themselves. Both methods can utilize the capability of IPv6 hosts to use the new address for new connections and to continue using the old address for existing connections. This capability to maintain two addresses ensures a smooth migration to a new network prefix.

Improve Scalability

You saw in Chapter 2 how IPv4 addresses restrict the scalability of internetworks. This section recaps those scalability problems. The first IPv4 problem is the limit of 32 bits for addressing, one of the main drivers for designing a new protocol. Pundits assumed that without intervention, IPv4 addresses would be depleted by the mid-1990s. That did not happen. NAT prolonged the life of IPv4 by allowing enterprises to use private addresses that are hidden from the public Internet. IPv6's 128 bits of address space allows many more globally routable devices to connect to the Internet. Private address space is also defined in IPv6.

Another problem with IPv4 is the large size of the Internet routing tables. CIDR was introduced to minimize the table size by introducing more hierarchy by aggregating addresses. However, many addresses cannot be aggregated. Addresses that were assigned before CIDR and addresses used by networks with certain multihomed Internet connections, for instance, cannot be aggregated.

IPv6 is designed for scalability, ease of configuration, and security, drawing from the lessons learned with IPv4. It is not designed to solve the Internet routing table size explosion. With strict allocation rules and procedures initiated from the start, and adherence to the hierarchy format for aggregation, however, table size can be contained. The goal is to achieve as much aggregation as possible, and the defined format of the globally routable address space facilitates this goal.

Ease of Configuration

IPv6 introduces mechanisms to ease host-to-router communication management and host configuration. These mechanisms are essential to the success of IPv6. As more and more people, schools, and businesses want to connect to the Internet or build their own internetworks, the tasks involved in enabling them must be simplified. Not everyone wants to become a CCIE just so he or she can figure out how to run a network. They just want the networks to work. IPv6 has automatic configuration mechanisms that enable hosts to obtain IP addresses, discover neighbors and default routers, and effectively use multiple default routers for redundancy.

Large companies connected to the Internet want the flexibility to change service providers without creating turmoil within their own networks. Renumbering networks will still be required with IPv6, but renumbering is made easier with the ability to maintain multiple addresses on all nodes and to have two different address states—one for use with active addresses and the other for use when an address is being phased out. In addition, network prefixes are advertised by routers to hosts, enabling the hosts to automatically configure themselves with IPv6 addresses. A company that needs to re-address its network because it changed ISPs can configure the routers to advertise the new prefixes as well as the old prefix. Hosts that receive the advertisement can automatically configure themselves with the new prefix information and can begin using the new addresses when new IP connections are made. Existing connections will continue to use the old address.

Security

People and businesses do not want to worry about security either. They want their data to be secure without thinking much about it. Authentication and encryption are built into IPv6. IPv6 packets can now be secured at the network layer within the network protocol.