Metropolitan area networks (MANs) and last-mile delivery—or access—networks are often designed using rings, as shown in Figure 12-5.
Figure 12-5 Common Last-Mile Technologies Access networks connect end users—organizations and individual users—to the global Internet.
Figure 12-5 shows
• A fixed wireless access (FWA) last mile. Connectivity to the towers or transmitters supporting individual homes is normally either a fiber ring or a hub-and-spoke topology.
• A fiber to the neighborhood (FTTN) or fiber to the curb (FTTC) network using GPON. Connections to the upstream routers are often a ring or hub-and-spoke topology.
• A DOCSIS, or cable television, connection. Cable television may be carried to neighborhood or curb using fiber technology in either a ring or hub-and-spoke topology.
• A MAN. Metropolitan Area Networks are normally built using ROADMs as a ring within a metropolitan area. Each ROADM may connect to a hub-and-spoke network, or another ring serving a smaller area like a corporate campus.
The last mile is often the hardest for service providers to support because
• The cost of installing cable to individual houses and businesses is generally very high.
• Subscription fees for individual users are generally low.
The cost of installing a cable to an individual might only be recouped after 10 or 15 years of service. Over 15 years, several new generations of networking technology can be invented, designed, and deployed. More sparsely populated areas are more difficult to support financially. Providers often only provide fixed wireless or satellite connectivity in sparsely populated areas.
Transit Provider Design
Transit providers connect last-mile providers, cloud providers, and other large-scale networks to build the core of the global Internet. Figure 12-6 illustrates part of one provider’s connectivity map—Lumen.
Figure 12-6 A Portion of Lumen Network’s Core
These are obviously large-scale, complex networks. Most of the time, however, even large-scale providers use interconnected rings to build their networks. There are few specializedtopologies used in WANs.
The dots on the network diagram shown in Figure 12-6 are regional provider facilities. Each dot may represent one or more physical facilities; more populated areas may have multiple physical facilities. There are several kinds of facilities a large-scale transit provider may operate, including
• Point of presence ( PoP): This is generally a facility used by the provider to terminate circuits.
• Colocation facility ( CoLo): A provider’s customer can rent space in a CoLo, allowing them to connect hosts directly to the provider’s core network. This allows the customer to place servers topologically closer to lots of users and allow the customer to pay the provider for power, space, control over physical access to the hosts, and network connectivity.
• Peering point: Providers connect to other providers and customers at peering points. These peering points can either be public or private. Anyone can rent space in a public peering point, install a router, and connect to the provider’s network.
Private peering points, on the other hand, can be accessed only by other providers, and are often managed under some form of cost-sharing arrangement between the connecting providers.
Providers connect to other providers for many reasons, but primarily to send traffic from one provider to another. There are three basic kinds of peering in the Internet:
• Provider-customer peering: When an organization wants to gain access to the Internet, it can pay a provider for a connection. Last-mile providers often pay global transit providers for upstream connectivity.
• Settlement-free peering: When two providers would like to connect to one another, and they anticipate the amount of traffic to be about the same in both directions, they can build a settlement-free peering connection.
• Open peering: Some organizations are given a connection to the Internet without payment (or settlements), even though the traffic levels are mismatched, or the organization would normally be considered a customer of the transit provider. For instance, research organizations are sometimes offered a connection at no cost to support the larger networking community. Social media networks are often also offered service at no cost because the transit provider would prefer the social media network’s user traffic stay on their network.
Large-scale transit providers tend to peer with a lot of other organizations.
Ring and Hub-and-Spoke Topologies The previous sections in this chapter described two basic network topologies: ring and hub-and-spoke. Let’s take a closer look at these two topologies. Figure 12-7 illustrates a network with ring and hub-and-spoke topologies.
Figure 12-7 Ring and Hub-and-spoke Topologies In Figure 12-7:
• Routers A through H are configured in a ring topology.
• Routers B and D, along with the three transmitter towers connected to them, are a dual-homed hub-and-spoke network topology.
• Routers E and G, along with the business buildings connected to them, are a dual-homed hub-and-spoke network topology.
• Router F and the three residences connected to it are a single-homed hub-and-spoke network topology.
The advantages of a ring topology are as follows:
• You can build large networks while using only a pair of interfaces on each network device.
• The ring is a two-connected topology, which means there are at least two paths through the network. If a single router or link fails, the network can still pass traffic to all the remaining destinations.
The primary disadvantage of the ring topology is the available bandwidth to reach any destination decreases for each connected node (router or switch) in the ring. If a new router is added to the ring between C and E, the C to E traffic must now compete with traffic entering the ring from this new router.
Each additional router adds more traffic without adding more overall bandwidth capacity. A second disadvantage of ring topologies is each additional node added to the ring causes a bit more delay and potentially adds a bit more jitter.
The primary advantage of the hub-and-spoke topology is the ability to support a lot of end locations, such as businesses or homes, with the minimal amount of wiring possible, particularly in densely populated areas. New destinations can be added without reducing the performance of existing destinations,.
The primary disadvantage of the single-homed hub-and-spoke topology is the loss of a single device or link disrupts service; every hub and link is a single point of failure. Adding a second hub device, and a second set of links, eliminates this single point of failure. Dual-homed hub-and-spoke networks are, however, a lot more complex than single-homed hub-and-spoke networks.