(UPDATED)
A Peering Agreement is usually a documentation trail that indicates how two network operators are going to interconnect. They are required when operators have Selective or Restricted peering policies.
The documentation trail can take a range of forms depending on the operators involved, and can range from email exchanges all the way to a formal contract between the two entities.
A peering agreement will usually contain the following information:
The AS numbers to be used by each party
The AS numbers of customers that are to be transited by each party (if applicable)
IP address space (IPv4 and/or IPv6) to be announced by each party (and by each party's customers as applicable)
Subnets of the IP address space to be announced by each party (acceptable ranges, used for traffic engineering)
The location of the physical interconnect
The bandwidth/capacity of the interconnect
Any specific BGP options (password to protect the session, use of MEDs, use of BGP communities,…)
Administrative contact for each party
Technical contact for each party
Network Operations Centre (24×7) contact for each party
What happens in case of network problems (escalation path) or disputes
Much of this is already contained in a well documented PeeringDB entry, but many operators still request a direct agreement so that there is supporting documentation indicating how the two parties will interact with each other.
Back to "What is required for Peering" page