In the early days of switched networks, switching was fast (often at hardware speed, meaning the speed was equivalent to the time it took to physically receive and forward frames onto other ports) and routing was slow (routing had to be processed in software). This prompted network designers to extend the switched portion of the network as much as possible. Access, distribution, and core layers were often configured to communicate at Layer 2. This topology created loop issues. To solve these issues, spanning-tree technologies were used to prevent loops while still enabling flexibility and redundancy in inter-switch connections.
However, as network technologies have evolved, routing has become faster and cheaper. Today, routing can be performed at hardware speed. One consequence of this evolution is that routing can be transferred to the core and the distribution layers without impacting network performance.
Many users are in separate VLANs, and each VLAN is usually a separate subnet. Therefore, it is logical to configure the distribution switches as Layer 3 gateways for the users of each access switch VLAN. This implies that each distribution switch must have IP addresses matching each access switch VLAN.
Layer 3 (routed) ports are normally implemented between the distribution and the core layer.
The network architecture depicted is not dependent on spanning tree because there are no physical loops in the Layer 2 portion of the topology.