This article relies largely or entirely on a single source. (June 2009)
In computer networking, split-horizon DNS, split-view DNS, split-brain DNS, or split DNS is the facility of a Domain Name System (DNS) implementation to provide different sets of DNS information, usually selected by the source address of the DNS request.
This facility can provide a mechanism for security and privacy management by logical or physical separation of DNS information for network-internal access (within an administrative domain, e.g., company) and access from an unsecure, public network (e.g. the Internet).
Implementation of split-horizon DNS can be accomplished with hardware-based separation or by software solutions. Hardware-based implementations run distinct DNS server devices for the desired access granularity within the networks involved. Software solutions use either multiple DNS server processes on the same hardware or special server software with the built-in capability of discriminating access to DNS zone records. The latter is a common feature of many server software implementations of the DNS protocol (cf. Comparison of DNS server software) and is sometimes the implied meaning of the term split-horizon DNS, since all other forms of implementation can be achieved with any DNS server software.
Split-Horizon DNS and DNSSEC
Split-horizon DNS is designed to provide different authoritative answers to an identical query and DNSSEC is used to ensure veracity of data returned by the Domain Name System. These apparently conflicting goals create the potential for confusion or false security alerts in poorly constructed networks. Research has produced recommendations to properly combine these two DNS features.
One common use case for split-horizon DNS is when a server, host1 in the example below, has both a private IP address on a local area network (not reachable from most of the Internet) and a public address, i.e. an address reachable across the Internet in general. By using split-horizon DNS the same name can lead to either the private IP address or the public one, depending on which client sends the query. This allows for critical local client machines to access a server directly through the local network, without the need to pass through a router. Passing through fewer network devices improves the network latency.
Internal DNS view, the server is named host1.example.net:
@ IN SOA ns.example.net admin.example.net. ( 2010010101 ; serial 1D ; refresh 1H ; retry 1W ; expire 3H ) ; minimum @ IN NS ns ns IN A 203.0.113.2 host1 IN A 10.0.0.10
@ IN SOA ns.example.net admin.example.net. ( 2010010101 ; serial 1D ; refresh 1H ; retry 1W ; expire 3H ) ; minimum @ IN NS ns ns IN A 203.0.113.2 host1 IN A 203.0.113.10
This article includes a list of references, but its sources remain unclear because it has insufficient inline citations. (June 2009) (Learn how and when to remove this template message)