Switches remember the address of every node on the network, and anticipate where data needs to go. Nodes connected to a switch can expect an immediate 40%-60% increase in performance.
A switch can also connect networks of different speeds together. A 100Mbps network, for example, could be connected to a slower 10Mbps network by inserting a switch between the two networks. In this way, switches are good for migrating to faster network speeds without having to discard older legacy network hardware.
In a telecommunications network, a switch is a device that channels incoming data from any of multiple input ports to the specific output port that will take the data toward its intended destination.
In the Open Systems Interconnection (OSI) communications model, a switch performs the layer 2 or Data-Link layer function. That is, it simply looks at each packet or data unit and determines from a physical address which device a data unit is intended for and switches it out toward that device. Some newer switches also perform routing functions, that is layer 3 or the Network layer functions in OSI and are sometimes called IP switches. .
On larger networks, the trip from one switch point to another in the network is called a hop. The time a switch takes to figure out where to forward a data unit is called its latency. The price paid for having the flexibility that switches provide in a network is this latency. Switches are found at the backbone and gateway levels of a network where one network connects with another and at the subnetwork level where data is being forwarded close to its destination or origin. The former are often known as core switches and the latter as desktop switches. .
In the simplest networks, a switch is not required for messages that are sent and received within the network. For example, a local area network may be organized in a token ring or bus arrangement in which each possible destination inspects each message and reads any message with its address.