circuit switching
Think of the old telephone system. When you placed a call, the network set up a dedicated path of wires and switches from your phone to the other phone, reserved just for you, for the whole length of the call — even during silences. That reserved end-to-end path is a circuit, and reserving one before any data flows is circuit switching.
Mechanically, circuit switching has three phases: setup (the network finds and reserves a path and a fixed slice of capacity end to end), transfer (data flows along that guaranteed path at the reserved rate, with predictable, constant delay and no queuing), and teardown (the reservation is released when you hang up). Because the slice is dedicated, you get a guaranteed rate and steady delay — wonderful for a phone call. Multiple circuits share a physical link by dividing it, for example into time slots (time-division multiplexing) or frequency bands (frequency-division multiplexing).
The catch is waste. A reserved circuit sits idle whenever you are not actively sending — and computer traffic is extremely bursty (a web click, then nothing, then a burst). Holding a fixed slice for a mostly-idle conversation squanders capacity, and setup adds delay before you can send anything. That is why the Internet uses packet switching instead. Circuit switching still appears where steady guaranteed rate matters (some optical and cellular transport, and the concept lives on in virtual circuits).
An old phone call held a circuit for its whole duration, so it never lagged or dropped words — but the same reserved path was wasted during every pause in the conversation.
Reserve first, then send — guaranteed and steady, but idle reservations waste capacity.
Circuit switching is not 'old and bad' — it gives guarantees packet switching cannot. The Internet chose packet switching because bursty computer traffic makes dedicated circuits mostly idle and thus wasteful.