Building Resilient IP Networks: The Access Module
- Multilayer Campus Design
- Access Module Building Blocks
- Layer 2 Domain
- Layer 3 Domain
- Summary
This chapter covers the following topics:
- Multilayer Campus Design
- Access Module Building Blocks
- Layer 2 Domain
- Layer 3 Domain
The access module is a network's interface to the users or end stations, and challenges in the access module are often related to the physical or Layer 2 connectivity problems. Many Layer 2 technologies can be used to build the access module; however, the focus of this chapter is on the Ethernet technology because Ethernet has emerged as the de facto connectivity standard for end devices.
The ubiquity of the Ethernet technology lies in its simplicity and cost-effectiveness. The availability of Layer 3 switching technology brings IP and Ethernet together, and this has a profound impact on IP network design. For the first time, it is possible to build a complete network for an entire company using IP+Ethernet strategy, including workstation connections, servers in the data center, and connecting branches via a metro Ethernet offering from a service provider. The integration between IP and Ethernet here is so tightly coupled that problems found in the Layer 2 network directly impact the overall IP network availability.
This chapter focuses on Ethernet switching technology, specifically Layer 2 network resiliency and how it should be built to provide a solid foundation for the Layer 3 network.
Multilayer Campus Design
With the popularity of Layer 3 switching, the term multilayer campus design is almost synonymous with the Ethernet switching design. A multilayer campus design has two main characteristics:
- Hierarchical— Each layer has a specific role to play.
- Modular— The entire network is built by piecing building blocks together.
These two characteristics enable the network to scale in a deterministic manner, with efficient use of resources to provide a resilient network foundation.
Figure 6-1 shows the concept of a typical multilayer campus design. It has an access layer, which commonly consists of wiring closet switches. The access layer is connected to a distribution layer. The distribution layer is, in turn, connected to the core layer. From a high-level view, a group of access switches are connected to a pair of distribution switches to form a basic building block. Many of these building blocks exist within the network, and the core layer connects them. Designs such as this are also known as three-tier architecture.
Figure 6-1 Three-Tier Multilayer Campus Design
Although Figure 6-1 shows a typical three-tier architecture, a smaller network may be built without the need of the core module at all, which makes it a two-tier design, as shown in Figure 6-2.
Figure 6-2 Two-Tier Multilayer Campus Design
The benefit of adopting the multilayer campus design is clarity of roles performed by each layer. The role of each layer is translated to a set of features required, which in turn translates to which particular type of hardware is to be used.
Access Layer
The access layer within the multilayer campus design model is where users gain access to the network. Most of the features found within the access layer are geared toward collecting and conditioning the traffic that is coming in from the users' end stations. These features include the following:
- Aggregating all the user endpoints
- Providing traffic conditioning functions such as marking and policing
- Providing intelligent network services such as automatic IP phone discovery
- Providing network security services such as 802.1x and port security
- Providing redundant links toward the distribution layer
In the classic multilayer campus design, the access layer is mainly made up of Layer 2 switches. Therefore, most of the work done here is in optimizing the Layer 2 protocol that governs this layer. This helps to provide a robust Layer 2 environment for the functioning of the IP network.
Distribution Layer
The distribution layer within the multilayer campus design model aggregates the access layer. One of the most important characteristics of the distribution layer is that it is the point where the Layer 2 domain ends and where the Layer 3 domain begins. The features at the distribution layer include the following:
- Aggregating access layer switches
- Terminating virtual LANs (VLANs) that are defined within the Layer 2 domain
- Providing the first-hop gateway service for all the end stations
- Providing traffic conditioning services such as security, quality of service (QoS), and queuing
- Providing redundant links toward the core layer, if required
Because the distribution layer is the meeting place for both the Layer 2 and Layer 3 domains, it runs both Layer 2 and Layer 3 protocols. This is also the place where most of the network intelligence is found and is perhaps one of the most complex parts of the network.
Core Layer
The core layer within the multilayer campus design model has two important tasks:
- Interconnect all the distribution layer blocks
- Forward all the traffic as quickly as it can
As the backbone of the entire network, its function is quite different from that of the access layer and distribution layer. The features that are critical to the functioning of the core layer include the following:
- Aggregating distribution layers to form an interconnected network
- Providing high-speed transfer of traffic among the distribution layers
- Providing a resilient IP routing environment
Because speed is of the essence here, the core layer usually does not provide services that may affect its performance (for example, security, access control, or any activities that require it to process every packet).
In the discussion of multilayer campus design, the inclusion of a core layer is always an interesting question. For a small network, it is common to see a two-tier design, as shown earlier in Figure 6-2, for cost reasons. However, for bigger networks, inclusion of the core layer is always recommended to scale the network in a manageable fashion.