Transit

Transit is the commercial service in which one autonomous system (the customer) pays another (the provider) for full internet reachability. The provider announces the customer's prefixes to the rest of the internet via BGP and, in return, delivers traffic from any destination back to the customer. Unlike peering, transit explicitly includes reachability to networks the customer has no direct relationship with.

In BGP terms, a transit provider sends the customer a full routing table (or a default route) and accepts the customer's prefixes for re-announcement to its own peers, transit providers and customers. This is what makes the customer's prefixes globally reachable. Networks usually contract with more than one transit provider to obtain redundancy, geographic diversity and the ability to engineer traffic by adjusting BGP attributes such as LOCAL_PREF or AS_PATH prepending.

The practical difference from peering is scope: a peer delivers traffic only to itself and its customers, a transit provider delivers traffic to everywhere it can reach, including via its own peers and upstream transits. Most operational networks therefore combine both — peering with as many networks as possible to reduce cost and latency, and one or more transit providers as the fallback path for everything else.

Commercially, the transit market is stratified. Tier-1 operators reach the entire default-free zone exclusively through settlement-free peering with other tier-1s and sell transit downstream; tier-2 and tier-3 operators buy transit upstream and either resell it or consume it themselves. Transit ports are typically billed per Mbit/s on the 95th percentile of measured traffic against a committed-information-rate, usually with a monthly minimum commit. That cost structure is precisely what motivates operators to shift traffic onto peering as soon as the volume toward a given destination justifies the effort.