This chapter is from the book
This chapter is from the book
This chapter is from the book
Access-Distribution Design Model
Chapter 2, “Enterprise Layer 2 and Layer 3 Design,” discussed different Layer 2 design models that are applicable to the campus LAN design, in particular to the access-distribution layer. Technically, each design model has different design attributes. Therefore, network designers must understand the characteristics of each design model to be able to choose and apply the most feasible model based on the design requirements.
The list that follows describes the three primary and common design models for the access layer to distribution layer connectivity. The main difference between these design models is where the Layer 2 and Layer 3 boundary is placed and how and where Layer 3 gateway services are handled:
- Classical multitier STP based: This model is the classical or traditional way of connecting access to the distribution layer in the campus network. In this model, the access layer switches usually operate in Layer 2 mode only, and the distribution layer switches operate in Layer 2 and Layer 3 modes. As discussed earlier in this book, the primary limitation of this design model is the reliance on Spanning Tree Protocol (STP) and First Hop Redundancy Protocol (FHRP). For more information, see Chapter 2.
Routed access: In this design model, access layer switches act as Layer 3 routing nodes, providing both Layer 2 and Layer 3 forwarding. In other words, the demarcation point between Layer 2 and Layer 3 is moved from the distribution layer to the access layer. Based on that, the Layer 2 trunk links from access to distribution are replaced with Layer 3 point-to-point routed links, as illustrated in Figure 3-5.
Figure 3-5 Routed Access Layer
The routed access design model has several advantages compared to the multitier classical STP-based access-distribution design model, including the following:
- Simpler and easier to troubleshoot, you can use a standard routing troubleshooting techniques, and you will have fewer protocols to manage and troubleshoot across the network
- Eliminate the reliance on STP and FHRP and rely on the equal-cost multipath (EMCP) of the used routing protocol to utilize all the available uplinks, which can increase the overall network performance
- Minimize convergence time during a link or node failure
Switch clustering: As discussed in Chapter 2, this design model provides the simplest and most flexible design compared to the other models discussed already. As illustrated in Figure 3-6, by introducing the switch clustering concept across the different functional modules of the enterprise campus architecture, network designers can simplify and enhance the design to a large degree. This offers a higher level of node and path resiliency, along with significantly optimized network convergence time.
Figure 3-6 Switch Clustering Concept
)
)
The left side of Figure 3-6 represents the physical connectivity, and the right side shows the logical view of this architecture, which is based on the switch clustering design model across the entire modular campus network.
Table 3-1 compares the different access-distribution connectivity design models from different design angles.
Table 3-1 Comparing Access-Distribution Connectivity Models
Multitier STP Based |
Routed Access |
Switch Clustering |
|
* Some switch clustering technologies, such as Cisco Nexus vPC, use FHRP (Hot Standby Router Protocol [HSRP]). However, from a forwarding plane point of view, both upstream switches (vPC peers) do forward traffic, unlike the -classical behavior, which is based on active-standby. |
|||
Design flexibility |
Limited (topology dependent) |
Limited (For example, spanning Layer 2 over different access switches requires an overlay technology) |
Flexible |
Scalability |
Supports scale up and limited scale out (topology dependent) |
Supports both scale up and scale out |
Scale up and limited scale out (typically limited to 2 distribution switches per cluster) |
Layer 3 gateway services |
Distribution layer (FHRP based) |
Access layer (Layer 3 routing based) |
Distribution layer (may or may not require FHRP*) |
Multichassis link aggregation (mLAG) |
Not supported |
Not supported (instead relies on Layer 3 ECMP) |
Supported |
Access-to-distribution convergence time |
Dependent on STP and FHRP timers (relatively slow) |
Interior Gateway Protocol (IGP) dependent, commonly fast |
Fast |
Operational complexity |
Complex (multiple control protocols to deal with [for example, STP, FHRP]) |
Moderate (Advanced routing design expertise may be required) |
Simple |