Two-Tier Campus Design (Collapsed Core)

To sift through all the requirements for a campus LAN, and then have a reasonable conversation about it with peers, most Cisco-oriented LAN designs use some common terminology to refer to the design. For this book’s purposes, you should be aware of some of the key campus LAN design terminology.

The Two-Tier Campus Design

Figure 13-1 shows a typical design of a large campus LAN, with the terminology included in the figure. This LAN has around 1000 PCs connected to switches that support around 25 ports each. Explanations of the terminology follow the figure.

Cisco uses three terms to describe the role of each switch in a campus design: access, distribution, and core. The roles differ based on whether the switch forwards traffic from user devices and the rest of the LAN (access), or whether the switch forwards traffic between other LAN switches (distribution and core).

Access switches connect directly to end users, providing user device access to the LAN. Access switches normally send traffic to and from the end-user devices to which they are connected and sit at the edge of the LAN.

Distribution switches provide a path through which the access switches can forward traffic to each other. By design, each of the access switches connects to at least one distribution switch, typically to two distribution switches for redundancy. The distribution switches provide the service of forwarding traffic to other parts of the LAN. Note that most designs use at least two uplinks to two different distribution switches (as shown in Figure 13-1) for redundancy.

The figure shows a two-tier design, with the tiers being the access tier (or layer) and the distribution tier (or layer). A two-tier design solves two major design needs:

• Provides a place to connect end-user devices (the access layer, with access switches)

• Connects the switches with a reasonable number of cables and switch ports by connecting all 40 access switches to two distribution switches

Note

The terms two-tier and 2-tier are synonyms, as are the terms three-tier and 3-tier. Cisco happens to use the versions of these terms with numerals in the exam topics.

Topology Terminology Seen Within a Two-Tier Design

The networking world uses several common terms about LAN and WAN topology and design including these:

Star: A design in which one central device connects to several others, so that if you drew the links out in all directions, the design would look like a star with light shining in all directions.

Full mesh: For any set of network nodes, a design that connects a link between each pair of nodes.

Partial mesh: For any set of network nodes, a design that connects a link between some pairs of nodes, but not all. In other words, a mesh that is not a full mesh.

Hybrid: A design that combines topology design concepts into a larger (typically more complex) design.

Armed with those formal definitions, note that the two-tier design is indeed a hybrid design that uses both a star topology at the access layer and a partial mesh at the distribution layer. To see why, consider Figure 13-2. It redraws a typical access layer switch, but instead of putting the PCs all below the switch, it spreads them around the switch. Then on the right, a similar version of the same drawing shows why the term star might be used—the topology looks a little like a child’s drawing of a star.

The distribution layer creates a partial mesh. If you view the access and distribution switches as nodes in a design, some nodes have a link between them, and some do not. Just refer to Figure 13-1 and note that, by design, none of the access layer switches connect to each other.

Finally, a design could use a full mesh. However, for a variety of reasons beyond the scope of the design discussion here, a campus design typically does not need to use the number of links and ports required by a full mesh design. However, just to make the point, first consider how many links and switch ports would be required for a single link between nodes in a full mesh, with six nodes, as shown in Figure 13-3.

Even with only six switches, a full mesh would consume 15 links (and 30 switch ports—two per link).

Now think about a full mesh at the distribution layer for a design like Figure 13-1, with 40 access switches and two distribution switches. Rather than drawing it and counting it, the number of links is calculated with this old math formula from high school: N(N – 1) / 2, or in this case, 42 * 41 / 2 = 861 links, and 1722 switch ports consumed among all switches.

For comparison’s sake, the partial mesh design of Figure 13-1, with a pair of links from each access switch to each distribution switch, requires only 160 links and a total of 320 ports among all switches.

Three-Tier Campus Design (Core)

The two-tier design of Figure 13-1, with a partial mesh of links at the distribution layer, happens to be the most common campus LAN design. It also goes by two common names: a two-tier design (for obvious reasons) and a collapsed core (for less obvious reasons). The term collapsed core refers to the fact that the two-tier design does not have a third tier, the core tier. This next topic examines a three-tier design that does have a core, for perspective.

Imagine your campus has just two or three buildings. Each building has a two-tier design inside the building, with a pair of distribution switches in each building and access switches spread around the building as needed. How would you connect the LANs in each building? Well, with just a few buildings, it makes sense to simply cable the distribution switches together, as shown in Figure 13-4.

The design in Figure 13-4 works well, and many companies use this design. Sometimes the center of the network uses a full mesh, sometimes a partial mesh, depending on the availability of cables between the buildings.

However, a design with a third tier (a core tier) saves on switch ports and on cables in larger designs. And note that with the links between buildings, the cables run outside, are often more expensive to install, and are almost always fiber cabling with more expensive switch ports, so conserving the number of cables used between buildings can help reduce costs.

A three-tier core design, unsurprisingly at this point, adds a few more switches (core switches), which provide one function: to connect the distribution switches. Figure 13-5 shows the migration of the Figure 13-4 collapsed core (that is, a design without a core) to a three-tier core design.

By using a core design, with a partial mesh of links in the core, you still provide connectivity to all parts of the LAN and to the routers that send packets over the WAN, just with fewer links between buildings.

The following list summarizes the terms that describe the roles of campus switches:

• Access: Provides a connection point (access) for end-user devices. Does not forward frames between two other access switches under normal circumstances.

• Distribution: Provides an aggregation point for access switches, providing connectivity to the rest of the devices in the LAN, forwarding frames between switches, but not connecting directly to end-user devices.

• Core: Aggregates distribution switches in very large campus LANs, providing very high forwarding rates for the larger volume of traffic due to the size of the network.

Topology Design Terminology

To close the discussion of Enterprise LAN topology, the next topic applies some of the generic topology terms to a typical two-tier design.

Consider Figure 13-6, which shows a few of the terms. First, on the left, drawings often show access switches with a series of cables, parallel to each other. However, the combinations of an access switch and its access links is often called a star topology. Why? Look at the redrawn access switch in the center of the figure, with the cables radiating out from the center. It does not look like a real star, but it looks a little like a child’s drawing of a star, hence the term star topology.

The right side of the figure repeats a typical two-tier design, focusing on the mesh of links between the access and distribution switches. Any group of nodes that connect with more links than a star topology is typically called a mesh. In this case, the mesh is a partial mesh, because not all nodes have a direct link between each other. A design that connects all nodes with a link would be a full mesh.

Real networks make use of these topology ideas, but often a network combines the ideas together. For instance, the right side of Figure 13-6 combines the star topology of the access layer with the partial mesh of the distribution layer. So you might hear these designs that combine concepts called a hybrid design.

PoE Basics

The family of standards that supply power goes by the general name Power over Ethernet (PoE). With PoE, some device, typically a LAN switch, acts as the Power Sourcing Equipment (PSE)—that is, the device that supplies DC power over the Ethernet UTP cable (as shown in Figure 13-9). A device that has the capability to be powered over the Ethernet cable, rather than by some other power connector on the device, is called the Powered Device (PD).

PoE has a great advantage for devices installed to locations that often do not have a preinstalled power cable or power output. For instance, wireless design places APs in a wide range across the ceiling of a floor (or story) in a building. Also, IP video cameras might be placed in the ceiling corners inside or at various outside locations. Instead of running new power and new network cables to support each device, a single Ethernet cable run can supply power to the device while allowing normal Ethernet communications over the same cable and same wire pairs.

PoE also helps in some less obvious practical ways because it supplies DC power over the Ethernet cable, so the device does not need an AC/DC converter. For instance, devices like laptops and IP phones use a power cord that includes a power brick—an AC-to-DC converter—which converts the AC power from the power outlet to the DC power needed by the device. PoE supplies DC current over the Ethernet cable. So, for an IP Phone, for instance, no more power cable and no more power brick cluttering the desk or taking up a power outlet.

PoE Operation

PoE must have a means to avoid harming the devices on the end of the circuit. Every electrical device can be harmed by receiving too much current into the device, which is why electricians install circuit breakers and why we use surge protectors. Applying power over an Ethernet cable could have the same effect, harming the device on the other end, if the device does not support PoE. So PoE must (and does) have processes in place to determine if PoE is needed, and for how much power, before applying any potentially harmful power levels to the circuit.

PoE, standardized by the IEEE, extends the same IEEE autonegotiation mechanisms. In fact, the mechanisms need to work before the PD has booted, because the PD needs power before it can boot and initialize. By using these IEEE autonegotiation messages and watching for the return signal levels, PoE can determine whether the device on the end of the cable requires power (that is, it is a PD) and how much power to supply. This list details the major steps:

Step 1. Do not supply power on a PoE-capable port unless negotiation identifies that the device needs power.

Step 2. Use Ethernet autonegotiation techniques, sending low power signals and monitoring the return signal, to determine the PoE power class, which determines how much power to supply to the device.

Step 3. If the device is identified as a PD, supply the power per the power class, which allows the device to boot.

Step 4. Monitor for changes to the power class, both with autonegotiation and listening for CDP and LLDP messages from the PD.

Step 5. If a new power class is identified, adjust the power level per that class.

The negotiation processes result in the PDs signaling how many watts of power they would like to receive from the PSE. Depending on the specific PoE standard, the PSE will then supply the power, either over two pairs or four pairs, as noted in Table 13-2.

Cisco has been developing products to use some form of PoE since around 2000. Cisco has often developed prestandard power capabilities, like its original Cisco Inline Power (ILP) feature. Over time, the IEEE has produced standards similar to Cisco’s power features, with Cisco supporting the standard version once completed. However, for the most part, the Cisco literature refers to the more common names in the first column of the table.

PoE and LAN Design

Most of the LAN switch features discussed in this book (and in CCNA 200-301 Official Cert Guide, Volume 1) exist as software features. Once you learn about a software feature, in some cases all you have to do is configure the feature and start using it. (In some cases, you might need to research and license the feature first.) Regardless, adding the feature takes little or no prior planning.

PoE does require some planning and engineering effort when designing a LAN, both when planning for the cable plant (both Ethernet and electrical), as well as when planning for new networking hardware. Planning with PoE in mind prepares the network to supply power to network devices, rather than reacting and missing opportunities to save money and time.

The following list includes some of the key points to consider when planning a LAN design that includes PoE:

• Powered Devices: Determine the types of devices and specific models, along with their power requirements.

• Power Requirements: Plan the numbers of different types of PDs to connect into each wiring closet to build a power budget. That power budget can then be processed to determine the amount of PoE power to make available through each switch.

• Switch Ports: Some switches support PoE standards on all ports, some on no ports, some on a subset of ports. Research the various switch models so that you purchase enough PoE-capable ports for the switches planned for each wiring closet.

• Switch Power Supplies: Without PoE, when purchasing a switch, you choose a power supply so that it delivers enough power to power the switch itself. With PoE, the switch acts as a distributor of electrical power, so the switch power supply must deliver many more watts than it needs to run the switch itself. You will need to create a power budget per switch, based on the number of connected PDs, and purchase power supplies to match those requirements.

• PoE Standards versus Actual: Consider the number of PoE switch ports needed, the standards they support, the standards supported by the PDs, and how much power they consume. For instance, a PD and a switch port may both support PoE+, which supports up to 30 watts supplied by the PSE. However, that powered device may need at most 9 watts to operate, so your power budget needs to reserve less power than the maximum for those devices.

Chapter Review

One key to doing well on the exams is to perform repetitive spaced review sessions. Review this chapter’s material using either the tools in the book or interactive tools for the same material found on the book’s companion website. Refer to the “Your Study Plan” element for more details. Table 13-3 outlines the key review elements and where you can find them. To better track your study progress, record when you completed these activities in the second column.