Designing the Core Layer

The backbone for campus connectivity and is the aggregation point for the other layers.

Should provide scalability, high availability, and fast convergence to the network.

• The core layer should scale easily.

• A high-speed environment should use hardware acceleration, if possible.

• The core should provide a high level of redundancy and adapt to changes quickly.

  • Core devices should be more reliable.

  • Accommodate failures by rerouting traffic and responding quickly to changes in the network topology.

• Implements scalable protocols and technologies.

• Provides alternate paths and load balancing.

• Packet manipulation should be avoided, such as checking access lists and filtering, which could slow down the switching of packets.

Not all campus implementations require a campus core.

The core and distribution layer functions can be combined at the distribution layer for a smaller campus.

Provide Alternate Paths

An additional link providing an alternate path to a second core switch from each distribution switch offers redundancy to support a single link or node failure.

Core Redundant Triangles

Without a Core Layer

The distribution layer switches need to be fully meshed.

Can be difficult to scale.

Increases the cabling requirements.

Routing complexity of a full-mesh design increases as new neighbors are added.

Can be used on small campuses with no perspective of growing.

Collapsed Core Layer Architecture

In smaller networks, the core and the distribution layer can be only one.

• Eliminates the need for extra switching hardware and simplifies network implementation.

However, eliminates the advantages of the multilayer architecture, specifically fault isolation.

Avoid Single Points of Failure

With a hierarchical design

In Distribution and Core Layers the single points of failure are easy to avoid with redundant links.

• Don't forget redundant power and cooling!

In Access Layer, all L2 switches are single points of failure (only) to the user connected to them.

• Solution 1, redundant backup hardware activated by a (proprietary) supervision mechanism to “replace” faulty equipment.

  • Copies full configuration and state to backup hardware.

• Solution 2, have multiple connections between each user terminal and different access switches.

  • Requires multiple network cards in user terminals and more plugs/wiring.

Avoid Too Much Redundancy

Increases:

• Routing complexity.

• The number of ports used.

• Wiring.

Optimal Redundancy