Network Working Group R. Rosenbaum Request for Comments: 1464 Digital Equipment Corporation May 1993
This memo defines an Experimental Protocol for the Internet community. Discussion and suggestions for improvement are requested. Please refer to the current edition of the "IAB Official Protocol Standards" for the standardization state and status of this protocol. Distribution of this memo is unlimited.
While the Domain Name System (DNS) [2,3] is generally used to store predefined types of information (e.g., addresses of hosts), it is possible to use it to store information that has not been previously classified.
This paper describes a simple means to associate arbitrary string information (ASCII text) with attributes that have not been defined by the DNS. It uses DNS TXT resource records to store the information. It requires no change to current DNS implementations.
The Domain Name System is designed to store information that has both a predefined type and structure. Examples include IP addresses of hosts and names of mail exchangers. It would be useful to take advantage of the widespread use and scaleability of the DNS to store information that has not been previously defined.
This paper proposes the use of the DNS TXT resource record (defined in STD 13, RFC 1035) to contain new types of information. The principal advantage of such an approach is that it requires no change to most existing DNS servers. It is not intended to replace the process by which new resource records are defined and implemented.
To store new types of information, the TXT record uses a structured format in its TXT-DATA field. The format consists of the attribute name followed by the value of the attribute. The name and value are separated by an equals sign (=).
For example, the following TXT records contain attributes specified in this fashion:
host.widgets.com IN TXT "printer=lpr5" sam.widgets.com IN TXT "favorite drink=orange juice"
The general syntax is:
<owner> <class> <ttl> TXT "<attribute name>=<attribute value>"
Any printable ASCII character is permitted for the attribute name. If an equals sign is embedded in the attribute name, it must be quoted with a preceding grave accent (or backquote: "`"). A backquote must also be quoted with an additional "`".
Attribute Name Matching Rules
The attribute name is considered case-insensitive. For example, a lookup of the attribute "Favorite Drink" would match a TXT record containing "favorite drink=Earl Grey tea".
During lookups, TXT records that do not contain an unquoted "=" are ignored. TXT records that seem to contain a null attribute name, that is, the TXT-DATA starts with the character "=", are also ignored.
Leading and trailing whitespace (spaces and tabs) in the attribute name are ignored unless they are quoted (with a "`"). For example, "abc" matches " abc<tab>" but does not match "` abc".
Note that most DNS server implementations require a backslash (\) or double quote (") in a text string to be quoted with a preceding backslash. Accent grave ("`") was chosen as a quoting character in this syntax to avoid confusion with "\" (and remove the need for confusing strings that include sequences like "\\\\").
All printable ASCII characters are permitted in the attribute value. No characters need to be quoted with a "`". In other words, the first unquoted equals sign in the TXT record is the name/value delimiter. All subsequent characters are part of the value.
Once again, note that in most implementations the backslash character is an active quoting character (and must, itself, be quoted).
All whitespace in the attribute value is returned to the requestor (it is up to the application to decide if it is significant.)
<sp> indicates a space character. Attribute Attribute Internal Form External Form Name Value (server to resolver) (TXT record) color blue color=blue "color=blue" equation a=4 equation=a=4 "equation=a=4" a=a true a`=a=true "a`=a=true" a\=a false a\`=a=false "a\\`=a=false" = \= `==\= "`==\\=" string "Cat" string="Cat" "string=\"Cat\"" string2 `abc` string2=``abc`` "string2=``abc``" novalue novalue= "novalue=" a b c d a b=c d "a b=c d" abc<sp> 123<sp> abc` =123<sp> "abc` =123 "
The attributes can be accessed by the standard resolver library, but it is recommended that a library routine designed specially for this attribute format be used. Such a routine might provide an analogue to gethostbyname:
getattributebyname(objectname, name of object attributename, name of attribute attributevalue, pointer to buffer attributevaluelen) length of buffer
This routine would remove all quoting characters before returning the information to the caller. A more complex routine could return attributes with multiple values, or several different attributes.
To permit ease of interoperability and to reduce the chance of naming conflicts, a registration process for well known attribute names might be established. This could be a periodically updated list of names and/or adherence to other name registration mechanisms such as published object identifiers.
This paper does not address attribute name registration.
Some DNS server implementations place limits on the size or number of
TXT records associated with a particular owner. Certain
implementations may not support TXT records at all.
 Stahl, M., "Domain Administrators Guide", RFC 1032, Network Information Center, SRI International, November 1987.
 Mockapetris, P., "Domain Names - Concepts and Facilities", STD 13, RFC 1034, USC/Information Sciences Institute, November 1987.
 Mockapetris, P., "Domain Names - Implementation and
Specification", STD 13, RFC 1035, USC/Information Sciences Institute, November 1987.
 Mockapetris, P., "DNS Encoding of Network Names and Other Types", RFC 1101, USC/Information Sciences Institute, April 1989.
Security issues are not discussed in this memo.
Digital Equipment Corporation
550 King Street, LKG2-2/Z7
Littleton, MA 01460-1289