What it is and the problem it solves
Peering is a voluntary, settlement-free interconnection between autonomous systems to exchange downstream user traffic. It solves the problem of costly, asymmetric transit by letting networks trade traffic directly — cutting costs, improving latency, and increasing resilience.
How it works
Peering works by physically interconnecting administratively separate autonomous systems and exchanging BGP routing information. It relies on tacit norms, not contracts — formal agreements occur in just 0.07% of cases. Traffic flows only between downstream users of each network. No money changes hands: each retains revenue from its own customers.
What works
Public peering at IXPs and private peering via direct links both deliver measurable cost reduction and performance gains where traffic volumes and symmetry align. The model sustains global end-to-end reachability — because every major network peers with enough others to route around failures or bottlenecks.
What does not
Peering does not guarantee performance, uptime, or fairness. It fails when mutual benefit ends — triggering depeering with no recourse. It does not scale automatically: multilateral peering via route servers introduces complexity and asymmetry. It is not universal — 0.02% of arrangements mislabelled as ‘peering’ involve settlement.
What it changes
Peering changes who pays for global connectivity. It shifts cost and control from centralised transit providers to distributed, self-organising networks. It enables regional models like ‘donut peering’, altering how traffic flows across geographies — but does not eliminate dependency on transit for full reachability.
Is it worth your time
Yes — if you operate or depend on internet infrastructure, peering directly affects your traffic costs, latency, redundancy, and control. If you are a developer, product manager, or policy analyst without operational responsibility for routing, it remains essential context but rarely demands direct intervention.
