primary, secondary, master, slave and redundancy

DNS administrators mention terms like primary, secondary, master, slave and redundancy when talking about dns services. But, what does it all mean?

Redundancy is the easiest to understand, and it is the only factor that matters to outside users.

Redundancy is the practice and requirement of providing authoritative dns services from more than one dns server for a dns zone. The requirement is two servers and the recommendation is three. The most common practice is two.

A further strong recommendation is that the servers be in different network and geographic spaces. The purpose of this requirement is the protection of dns service availability for a dns zone in the event of network or natural disasters. It is often further recommended that the servers be placed in diverse autonomous network systems for greater diversity of routes.

The remaining terms, primary, secondary, master, and slave have meaning only with respect to internal dns administration.

A slave dns server is a dns server which maintains no independent dns zone data. Instead, it retrieves, or receives zone record data from one or more designated masters.

A master dns server is a dns server which maintains and stores authoritative dns zone data independently. This data may or may not be provided to other dns servers. A special case is the use of hidden masters. A hidden master is a master dns server which provides data only to other dns servers that are authoritative for the zone. This configuration is used as an administrative convenience.

Formerly, the analogous terms were primary dns servers and secondary dns servers. The only significance of the primary dns server is that it is the dns server which is associated with the start of authority record for the zone. Secondary dns servers are any dns servers which are not the dns server listed in the SOA record. Therefore, a secondary server can be a master or a slave.