This chapter includes the following sections:
- Network Hierarchy
- Using a Modular Approach to Network Design
- Services Within Modular Networks
- Network Management Protocols and Features
- Summary
- References
- Case Study: ACMC Hospital Modularity
- Review Questions
This chapter introduces a modular hierarchical approach to network design, the Cisco Enterprise Architecture. The chapter begins with a discussion of the hierarchical network structure. The next section introduces network modularization and discusses the details of the Cisco Enterprise Architecture. Following that are a detailed description of services within modular networks, and a discussion of network management protocols and features.
Network Hierarchy
This section explains the hierarchical network model, which is composed of the access, distribution, and core layers. The functions generally associated with each of these layers are discussed, as is the most common approach to designing a hierarchical network.
Historically used in the design of enterprise local-area network and wide-area network data networks, this model works equally well within the functional modules of the Cisco Enterprise Architecture. These modules are discussed later in this chapter, in the section "Using a Modular Approach to Network Design."
Hierarchical Network Model
The hierarchical network model provides a framework that network designers can use to help ensure that the network is flexible and easy to implement and troubleshoot.
Hierarchical Network Design Layers
As shown in Figure 3-1, the hierarchical network design model consists of three layers:
- The access layer provides local and remote workgroup or user access to the network.
- The distribution layer provides policy-based connectivity.
- The core (or backbone) layer provides high-speed transport to satisfy the connectivity and transport needs of the distribution layer devices.
)
Figure 3-1 Hierarchical Model's Three Layers
Each hierarchical layer focuses on specific functions, thereby allowing the network designer to choose the right systems and features based on their function within the model. This approach helps provide more accurate capacity planning and minimize total costs. Figure 3-2 illustrates a sample network showing the mapping to the hierarchical model's three layers.
)
Figure 3-2 Sample Network Designed Using the Hierarchical Model
You do not have to implement the hierarchical layers as distinct physical entities; they are defined to aid successful network design and to represent functionality that must exist within a network. The actual manner in which you implement the layers depends on the needs of the network you are designing. Each layer can be implemented in routers or switches, represented by physical media, or combined in a single device. A particular layer can be omitted, but hierarchy should be maintained for optimum performance. The following sections detail the functionality of the three layers and the devices used to implement them.
Access Layer Functionality
This section describes the access layer functions and the interaction of the access layer with the distribution layer and local or remote users.
The Role of the Access Layer
The access layer is the concentration point at which clients access the network. Access layer devices control traffic by localizing service requests to the access media.
The purpose of the access layer is to grant user access to network resources. Following are the access layer's characteristics:
- In the campus environment, the access layer typically incorporates switched LAN devices with ports that provide connectivity for workstations and servers.
- In the WAN environment, the access layer for teleworkers or remote sites provides access to the corporate network across some wide-area technology, such as Frame Relay, Multiprotocol Label Switching (MPLS), Integrated Services Digital Network, leased lines, Digital Subscriber Line (DSL) over traditional telephone copper lines, or coaxial cable.
- So as not to compromise network integrity, access is granted only to authenticated users or devices (such as those with physical address or logical name authentication). For example, the devices at the access layer must detect whether a telecommuter who is dialing in is legitimate, yet they must require minimal authentication steps for the telecommuter.
Layer 2 and Multilayer Switching in the Access Layer
Access can be provided to end users as part of either a Layer 2 (L2) switching environment or a multilayer switching environment.
Using Layer 2 Switching in the Access Layer
Access to local workstations and servers can be provided using shared or switched media LANs; VLANs may be used to segment the switched LANs. Each LAN or VLAN is a single broadcast domain.
The access layer aggregates end-user switched 10/100 ports and provides Fast Ethernet, Fast EtherChannel, and Gigabit Ethernet uplinks to the distribution layer to satisfy connectivity requirements and reduce the size of the broadcast domains. You can deploy multiple VLANs, each with its own IP subnet and its own instance of Spanning Tree Protocol (STP) providing alternative paths in case of failure. In this case, Layer 2 trunking (typically using the Institute for Electrical and Electronic Engineers [IEEE] 802.1Q trunking protocol) is used between the access layer switches and the distribution layer switches, with per-VLAN STP on each uplink for load balancing and redundancy, and with a distribution layer multilayer switch providing the inter-VLAN communication for the access layer.
When RSTP cannot be implemented, Cisco IOS STP features such as UplinkFast, PortFast, and BackboneFast can be used to provide equivalent convergence improvements. These features are described as follows:
- UplinkFast: Enables faster failover on an access layer switch on which dual uplinks connect to the distribution layer. The failover time is reduced by unblocking the blocked uplink port on a switch immediately after root port failure, thereby transitioning it to the forwarding state immediately, without transitioning the port through the listening and learning states.
- BackboneFast: If a link fails on the way to the root switch but is not directly connected to the local switch, BackboneFast reduces the convergence time from 50 seconds to between 20 and 30 seconds.
- PortFast: Enables switch ports connected to nonswitch devices (such as workstations) to immediately enter the spanning-tree forwarding state, thereby bypassing the listening and learning states, when they come up. Ports connected only to an end-user device do not have bridging loops, so it is safe to go directly to the forwarding state, significantly reducing the time it takes before the port is usable.
Using Multilayer Switching in the Access Layer
The most common design for remote users is to use multilayer switches or routers. A multilayer switch, or router, is the boundary for broadcast domains and is necessary for communicating between broadcast domains (including VLANs). Access routers provide access to remote office environments using various wide-area technologies combined with multilayer features, such as route propagation, packet filtering, authentication, security, Quality of Service (QoS), and so on. These technologies allow the network to be optimized to satisfy a particular user's needs. In a dialup connection environment, dial-on-demand routing (DDR) and static routing can be used to control costs.
Access Layer Example
Figure 3-3 illustrates a sample network in which the campus access layer aggregates end users and provides uplinks to the distribution layer. The access layer switches are dual-attached to the distribution layer switches for high availability.
)
Figure 3-3 Access Layer Connectivity in a Campus LAN
The access layer can support convergence, high availability, security, QoS, and IP multicast. Some services found at the access layer include establishing a QoS trust boundary, broadcast suppression, and Internet Group Management Protocol (IGMP) snooping.
Distribution Layer Functionality
This section describes distribution layer functions and the interaction of the distribution layer with the core and access layers.
The Role of the Distribution Layer
The distribution layer represents both a separation between the access and core layers and a connection point between the diverse access sites and the core layer. The distribution layer determines department or workgroup access and provides policy-based connectivity.
Following are the characteristics of the distribution layer:
- Distribution layer devices control access to resources that are available at the core layer and must therefore use bandwidth efficiently.
- In a campus environment, the distribution layer aggregates wiring closet bandwidth by concentrating multiple low-speed access links into a high-speed core link and using switches to segment workgroups and isolate network problems to prevent them from affecting the core layer.
- Similarly, in a WAN environment, the distribution layer aggregates WAN connections at the edge of the campus and provides policy-based connectivity.
- This layer provides redundant connections for access devices. Redundant connections also provide the opportunity to load-balance between devices.
- The distribution layer represents a routing boundary between the access and core layers and is where routing and packet manipulation are performed.
- The distribution layer allows the core layer to connect diverse sites while maintaining high performance. To maintain good performance in the core, the distribution layer can redistribute between bandwidth-intensive access-layer routing protocols and optimized core routing protocols. Route filtering is also implemented at the distribution layer.
- The distribution layer can summarize routes from the access layer to improve routing protocol performance. For some networks, the distribution layer offers a default route to access-layer routers and runs dynamic routing protocols only when communicating with core routers.
- The distribution layer connects network services to the access layer and implements policies for QoS, security, traffic loading, and routing. For example, the distribution layer addresses different protocols' QoS needs by implementing policy-based traffic control to isolate backbone and local environments. Policy-based traffic control prioritizes traffic to ensure the best performance for the most time-critical and time-dependent applications.
- The distribution layer is often the layer that terminates access layer VLANs (broadcast domains); however, this can also be done at the access layer.
- This layer provides any media transitions (for example, between Ethernet and ATM) that must occur.
Distribution Layer Example
Figure 3-4 shows a sample network with various features of the distribution layer highlighted.
)
Figure 3-4 Example of Distribution Layer Features
Following are the characteristics of the distribution layer in the routed campus network shown in Figure 3-4:
- Multilayer switching is used toward the access layer (and, in this case, within the access layer).
- Multilayer switching is performed in the distribution layer and extended toward the core layer.
- The distribution layer performs two-way route redistribution to exchange the routes between the Routing Information Protocol version 2 (RIPv2) and Enhanced Interior Gateway Routing Protocol (EIGRP) routing processes.
- Route filtering is configured on the interfaces toward the access layer.
- Route summarization is configured on the interfaces toward the core layer.
- The distribution layer contains highly redundant connectivity, both toward the access layer and toward the core layer.
Core Layer Functionality
This section describes core layer functions and the interaction of the core layer with the distribution layer.
The Role of the Core Layer
The function of the core layer is to provide fast and efficient data transport. Characteristics of the core layer include the following:
- The core layer is a high-speed backbone that should be designed to switch packets as quickly as possible to optimize communication transport within the network.
- Because the core is critical for connectivity, core layer devices are expected to provide a high level of availability and reliability. A fault-tolerant network design ensures that failures do not have a major impact on network connectivity. The core must be able to accommodate failures by rerouting traffic and responding quickly to changes in network topology. The core must provide a high level of redundancy. A full mesh is strongly suggested, and at least a well-connected partial mesh with multiple paths from each device is required.
- The core layer should not perform any packet manipulation, such as checking access lists or filtering, which would slow down the switching of packets.
- The core layer must be manageable.
- The core devices must be able to implement scalable protocols and technologies, and provide alternative paths and load balancing.
Switching in the Core Layer
Layer 2 switching or multilayer switching (routing) can be used in the core layer. Because core devices are responsible for accommodating failures by rerouting traffic and responding quickly to network topology changes, and because performance for routing in the core with a multilayer switch incurs no cost, most implementations have multilayer switching in the core layer. The core layer can then more readily implement scalable protocols and technologies, and provide alternate paths and load balancing.
Figure 3-5 shows an example of Layer 2 switching in the campus core.
)
Figure 3-5 Layer 2 Switching in the Campus Core
In Figure 3-5, a typical packet between access sites follows these steps:
- Step 1 The packet is Layer 2–switched toward a distribution switch.
- Step 2 The distribution switch performs multilayer switching toward a core interface.
- Step 3 The packet is Layer 2–switched across the LAN core.
- Step 4 The receiving distribution switch performs multilayer switching toward an access layer LAN.
- Step 5 The packet is Layer 2–switched across the access layer LAN to the destination host.
Figure 3-6 shows an example of multilayer switching in the campus core.
)
Figure 3-6 Multilayer Switching in the Campus Core
In Figure 3-6, a typical packet between access sites follows these steps:
- Step 1 The packet is Layer 2–switched toward a distribution switch.
- Step 2 The distribution switch performs multilayer switching toward a core interface.
- Step 3 The packet is multilayer-switched across the LAN core.
- Step 4 The receiving distribution switch performs multilayer switching toward an access LAN.
- Step 5 The packet is Layer 2–switched across the access layer LAN to the destination host.
Hierarchical Routing in the WAN
Figure 3-7 shows an example of hierarchical routing in the WAN portion of a network.
)
Figure 3-7 Hierarchical Routing in the WAN
In Figure 3-7, a typical packet between access sites follows these steps:
- Step 1 The packet is Layer 3–forwarded toward the distribution router.
- Step 2 The distribution router forwards the packet toward a core interface.
- Step 3 The packet is forwarded across the WAN core.
- Step 4 The receiving distribution router forwards the packet toward the appropriate access layer router.
- Step 5 The packet is Layer 3–forwarded to the destination host's access layer LAN.