Core-edge topology devices connect to edge switches, which are connected to each other through core switches. The example shown in the following figure uses three core switches. You could use more or fewer switches in the core, depending on whether you need higher availably and greater throughput, or a more efficient use of links and ports.
This topology is reliable, fast, and scales well. It is reliable because it has multiple core switches. If a core switch or a link to a core switch fails, an alternate path is available. As you increase the number of core switches, you also increase the number of link or core switch failures your cluster can tolerate.
High performance and low latency are ensured because throughput is high and the hop count is low. Throughput is high because multiple core switches share the load. Two hops get you from any edge switch to any other edge switch. If you need greater throughput, simply add another core switch.
Scaling the topology also requires additional core switches and links. However, the number of additional links you need is typically not as great as with, for example, a full mesh topology.