Good network design follows long-established principles like hierarchical design and information hiding. Basic network protocols exhibit this through data encapsulation. Lower layers hide their operation from the upper layers, allowing changes in the lower layers to not impact the upper layers (i.e. applications).
Network design recommendations are to use a Core, Distribution, and Access layer design. A concept that’s overlooked in this type of design is that one goal of the distribution layer is to prevent changes in the access layer (where end stations connect to the network) from affecting the core. The core should be designed to operate at maximum packet forwarding efficiency so as to provide high reliability, low latency, low delay, and low packet loss. In a sense, the distribution layer operates much like OSPF areas do – summarize changes in the access layer and only present to the core the summary information. Said another way, the distribution layer insulates the core from changes that occur at the access layer, allowing the core to be more stable and efficient.
I had not seen this idea specifically presented until a few weeks ago at a Cisco University presentation in Columbia, MD, and I thought it was worth sharing with a wider audience.
Re-posted with Permission
NetCraftsmen would like to acknowledge Infoblox for their permission to re-post this article which originally appeared in the Applied Infrastructure blog under http://www.infoblox.com/en/communities/blogs.html