HomeDocsIETF standardsRFCRFC 7500 — 7599
Text version
Internet Engineering Task Force (IETF)
Request for Comments: 7542
Obsoletes: 4282
Category: Standards Track
ISSN: 2070-1721
A. DeKok
FreeRADIUS
May 2015

The Network Access Identifier

Abstract

In order to provide inter-domain authentication services, it is necessary to have a standardized method that domains can use to identify each other's users. This document defines the syntax for the Network Access Identifier (NAI), the user identifier submitted by the client prior to accessing resources. This document is a revised version of RFC 4282. It addresses issues with international character sets and makes a number of other corrections to RFC 4282.

Status of This Memo

This is an Internet Standards Track document.

This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 5741.

Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc7542.

Copyright Notice

Copyright © 2015 IETF Trust and the persons identified as the document authors. All rights reserved.

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.

This document may contain material from IETF Documents or IETF Contributions published or made publicly available before November 10, 2008. The person(s) controlling the copyright in some of this material may not have granted the IETF Trust the right to allow modifications of such material outside the IETF Standards Process. Without obtaining an adequate license from the person(s) controlling the copyright in such materials, this document may not be modified outside the IETF Standards Process, and derivative works of it may not be created outside the IETF Standards Process, except to format it for publication as an RFC or to translate it into languages other than English.

Table of Contents

   1. Introduction ....................................................4
      1.1. Terminology ................................................6
      1.2. Requirements Language ......................................7
      1.3. Purpose ....................................................7
      1.4. Motivation .................................................9
   2. NAI Definition .................................................10
      2.1. UTF-8 Syntax and Normalization ............................10
      2.2. Formal Syntax .............................................11
      2.3. NAI Length Considerations .................................11
      2.4. Support for Username Privacy ..............................12
      2.5. International Character Sets ..............................13
      2.6. The Normalization Process .................................14
           2.6.1. Issues with the Normalization Process ..............15
      2.7. Use in Other Protocols ....................................16
      2.8. Using the NAI Format for Other Identifiers ................17
   3. Routing inside of AAA Systems ..................................18
      3.1. Compatibility with Email Usernames ........................19
      3.2. Compatibility with DNS ....................................20
      3.3. Realm Construction ........................................20
           3.3.1. Historical Practices ...............................21
      3.4. Examples ..................................................22
   4. Security Considerations ........................................23
      4.1. Correlation of Identities over Time and Protocols .........23
      4.2. Multiple Identifiers ......................................24
   5. Administration of Names ........................................25
   6. References .....................................................26
      6.1. Normative References ......................................26
      6.2. Informative References ....................................26
   Appendix A. Changes from RFC 4282 .................................29
   Acknowledgments ...................................................30
   Author's Address ..................................................30

1. Introduction

Considerable interest exists for a set of features that fit within the general category of inter-domain authentication, or "roaming capability" for network access, including dialup Internet users, Virtual Private Network (VPN) usage, wireless LAN authentication, and other applications.

By "inter-domain authentication", this document refers to situations where a user has authentication credentials at one "home" domain but is able to present them at a second "visited" domain to access certain services at the visited domain. The two domains generally have a pre-existing relationship, so that the credentials can be passed from the visited domain to the home domain for verification. The home domain typically responds with a permit/deny response, which may also include authorization parameters that the visited domain is expected to enforce on the user.

That is, the "roaming" scenario involves a user visiting, or "roaming" to, a non-home domain and requesting the use of services at that visited domain.

Interested parties have included the following:

  • Regional Internet Service Providers (ISPs) operating within a particular state or province, looking to combine their efforts with those of other regional providers to offer dialup service over a wider area.
  • Telecommunications companies who wish to combine their operations with those of one or more companies in other areas or nations, in order to offer more comprehensive network access service in areas where there is no native service (e.g., in another country).
  • Wireless LAN hotspots providing service to one or more ISPs.
  • Businesses desiring to offer their employees a comprehensive package of dialup services on a global basis. Those services may include Internet access as well as secure access to corporate intranets via a VPN, enabled by tunneling protocols such as the Point-to-Point Tunneling Protocol (PPTP) [RFC2637], the Layer 2 Forwarding (L2F) protocol [RFC2341], the Layer 2 Tunneling Protocol (L2TP) [RFC2661], and the IPsec tunnel mode [RFC4301].
  • Other protocols that are interested in leveraging the users' credentials in order to take advantage of an existing authentication framework.

In order to enhance the interoperability of these services, it is necessary to have a standardized method for identifying users. This document defines syntax for the Network Access Identifier (NAI). Examples of implementations that use the NAI, and descriptions of its semantics, can be found in [RFC2194].

When the NAI was defined for network access, it had the side effect of defining an identifier that could be used in non-AAA systems. Some non-AAA systems defined identifiers that were compatible with the NAI, and deployments used the NAI. This process simplified the management of credentials, by reusing the same credential in multiple situations. Protocols that reuse the same credential or the same identifier format can benefit from this simplified management. The alternative is to have protocol-specific credentials or identifier formats, which increases cost to both the user and the administrator.

There are privacy implications to using one identifier across multiple protocols. See Sections 2.7 and 4 for further discussion of this topic.

The goal of this document is to define the format of an identifier that can be used in many protocols. A protocol may transport an encoded version of the NAI (e.g., '.' as %2E). However, the definition of the NAI is protocol independent. The goal of this document is to encourage the widespread adoption of the NAI format. This adoption will decrease the work required to leverage identification and authentication in other protocols. It will also decrease the complexity of non-AAA systems for end users and administrators.

This document only suggests that the NAI format be used; it does not require such use. Many protocols already define their own identifier formats. Some of these are incompatible with the NAI, while others allow the NAI in addition to non-NAI identifiers. The definition of the NAI in this document has no requirements on protocol specifications, implementations, or deployments.

However, this document suggests that using one standard identifier format is preferable to using multiple incompatible identifier formats. Where identifiers need to be used in new protocols and/or specifications, it is RECOMMENDED that the format of the NAI be used. That is, the interpretation of the identifier is context specific, while the format of the identifier remains the same. These issues are discussed in more detail in Section 2.8, below.

The recommendation for a standard identifier format is not a recommendation that each user have one universal identifier. In contrast, this document allows for the use of multiple identifiers and recommends the use of anonymous identifiers where those identifiers are publicly visible.

This document is a revised version of [RFC4282], which originally defined internationalized NAIs. Differences and enhancements compared to that document are listed in Appendix A.

1.1. Terminology

This document frequently uses the following terms:

"Local" or "Localized" Text

"Local" or "localized" text is text that is in either non-UTF-8 or non-normalized form. The character set, encoding, and locale are (in general) unknown to Authentication, Authorization, and Accounting (AAA) network protocols. The client that "knows" the locale may have a different concept of this text than other AAA entities, which do not know the same locale. 

   Network Access Identifier

The Network Access Identifier (NAI) is a common format for user identifiers submitted by a client during authentication. The purpose of the NAI is to allow a user to be associated with an account name, as well as to assist in the routing of the authentication request across multiple domains. Please note that the NAI may not necessarily be the same as the user's email address or the user identifier submitted in an application-layer authentication. 

   Network Access Server

The Network Access Server (NAS) is the device that clients connect to in order to get access to the network. In PPTP terminology, this is referred to as the PPTP Access Concentrator (PAC), and in L2TP terminology, it is referred to as the L2TP Access Concentrator (LAC). In IEEE 802.11, it is referred to as an Access Point. 

   Roaming Capability

Roaming capability can be loosely defined as the ability to use any one of multiple Internet Service Providers (ISPs), while maintaining a formal customer-vendor relationship with only one. Examples of cases where roaming capability might be required include ISP "confederations" and ISP-provided corporate network access support. 

   Normalization or Canonicalization

These terms are defined in Section 4 of [RFC6365]; those definitions are incorporated here by reference. 

   Locale

This term is defined in [RFC6365], Section 8; that definition is incorporated here by reference. 

   Tunneling Service

A tunneling service is any network service enabled by tunneling protocols such as PPTP, L2F, L2TP, and IPsec tunnel mode. One example of a tunneling service is secure access to corporate intranets via a Virtual Private Network (VPN).

1.2. Requirements Language

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119].

1.3. Purpose

As described in [RFC2194], there are a number of providers offering network access services, and essentially all Internet Service Providers are involved in roaming consortia.

In order to be able to offer roaming capability, one of the requirements is to be able to identify the user's home authentication server. For use in roaming, this function is accomplished via the Network Access Identifier (NAI) submitted by the user to the NAS in the initial network authentication. It is also expected that NASes will use the NAI as part of the process of opening a new tunnel, in order to determine the tunnel endpoint.

This document suggests that other protocols can take advantage of the NAI format. Many protocols include authentication capabilities, including defining their own identifier formats. These identifiers can then end up being transported in AAA protocols, so that the originating protocols can leverage AAA for user authentication. There is therefore a need for a definition of a user identifier that can be used in multiple protocols.

While the NAI is defined herein, it should be noted that existing protocols and deployments do not always use it. AAA systems MUST therefore be able to handle user identifiers that are not in the NAI format. The process by which that is done is outside of the scope of this document.

Non-AAA systems can accept user identifiers in forms other than the NAI. This specification does not forbid that practice. It only codifies the format and interpretation of the NAI. This document cannot change existing protocols or practices. It can, however, suggest that using a consistent form for a user identifier is of benefit to the community.

This document does not make any protocol-specific definitions for an identifier format, and it does not make changes to any existing protocol. Instead, it defines a protocol-independent form for the NAI. It is hoped that the NAI is a user identifier that can be used in multiple protocols.

Using a common identifier format simplifies protocols requiring authentication, as they no longer need to specify a protocol-specific format for user identifiers. It increases security, as multiple identifier formats allow attackers to make contradictory claims without being detected (see Section 4.2 for further discussion of this topic). It simplifies deployments, as a user can have one identifier in multiple contexts, which allows them to be uniquely identified, so long as that identifier is itself protected against unauthorized access.

In short, having a standard is better than having no standard at all.

1.4. Motivation

The changes from [RFC4282] are listed in detail in Appendix A. However, some additional discussion is appropriate to motivate those changes.

The motivation to revise [RFC4282] began with internationalization concerns raised in the context of [EDUROAM]. Section 2.1 of [RFC4282] defines ABNF for realms and limits the realm grammar to English letters, digits, and the hyphen "-" character. The intent appears to have been to encode, compare, and transport realms with the Punycode [RFC3492] encoding form as described in [RFC5891]. There are a number of problems with this approach:

  • The [RFC4282] ABNF is not aligned with internationalization of DNS.
  • The requirement in Section 2.1 of [RFC4282] that realms are ASCII conflicts with the Extensible Authentication Protocol (EAP) as defined in [RFC3748], and RADIUS, which are both 8-bit clean, and which both recommend the use of UTF-8 for identifiers.
  • Section 2.4 of [RFC4282] required mappings that are language specific and that are nearly impossible for intermediate nodes to perform correctly without information about that language.
  • Section 2.4 of [RFC4282] requires normalization of usernames, which may conflict with local system or administrative requirements.
  • The recommendations in Section 2.4 of [RFC4282] for treatment of bidirectional characters have proven to be unworkable.
  • The prohibition of the use of unassigned code points in Section 2.4 of [RFC4282] effectively prohibits support for new scripts.
  • No Authentication, Authorization, and Accounting (AAA) client, proxy, or server has implemented any of the requirements in Section 2.4 of [RFC4282], among other sections.

With international roaming growing in popularity, it is important for these issues to be corrected in order to provide robust and interoperable network services.

Furthermore, this document was motivated by a desire to codify existing practice related to the use of the NAI format and to encourage widespread use of the format.

2. NAI Definition

2.1. UTF-8 Syntax and Normalization

UTF-8 characters can be defined in terms of octets using the following ABNF [RFC5234], taken from [RFC3629]: 

   UTF8-xtra-char  =   UTF8-2 / UTF8-3 / UTF8-4
   
   UTF8-2          =   %xC2-DF UTF8-tail
   
   UTF8-3          =   %xE0 %xA0-BF UTF8-tail /
                       %xE1-EC 2( UTF8-tail ) /
                       %xED %x80-9F UTF8-tail /
                       %xEE-EF 2( UTF8-tail )
   
   UTF8-4          =   %xF0 %x90-BF 2( UTF8-tail ) /
                       %xF1-F3 3( UTF8-tail ) /
                       %xF4 %x80-8F 2( UTF8-tail )
   
   UTF8-tail       =   %x80-BF

These are normatively defined in [RFC3629] but are repeated in this document for reasons of convenience.

See [RFC5198] and Section 2.6 of this specification for a discussion of normalization. Strings that are not Normal Form Composed (NFC) are not valid NAIs and SHOULD NOT be treated as such. Implementations that expect to receive an NAI but that instead receive non-normalized (but otherwise valid) UTF-8 strings instead SHOULD attempt to create a local version of the NAI, which is normalized from the input identifier. This local version can then be used for local processing. This local version of the identifier MUST NOT be used outside of the local context.

Where protocols carry identifiers that are expected to be transported over a AAA protocol, it is RECOMMENDED that the identifiers be in NAI format. Where the identifiers are not in the NAI format, it is up to the AAA systems to discover this and to process them. This document does not suggest how that is done. However, existing practice indicates that it is possible.

As internationalized domain names become more widely used, existing practices are likely to become inadequate. This document therefore defines the NAI, which is a user identifier format that can correctly deal with internationalized identifiers.

2.2. Formal Syntax

The grammar for the NAI is given below, described in Augmented Backus-Naur Form (ABNF) as documented in [RFC5234]. 

   nai            =   utf8-username
   nai            =/  "@" utf8-realm
   nai            =/  utf8-username "@" utf8-realm
   
   utf8-username  =  dot-string
   
   dot-string     = string *("." string)
   string         = 1*utf8-atext
   
   utf8-atext     =  ALPHA / DIGIT /
                     "!" / "#" /
                     "$" / "%" /
                     "&" / "'" /
                     "*" / "+" /
                     "-" / "/" /
                     "=" / "?" /
                     "^" / "_" /
                     "`" / "{" /
                     "|" / "}" /
                     "~" /
                     UTF8-xtra-char
   
   utf8-realm     =  1*( label "." ) label
   
   label          =  utf8-rtext *(ldh-str)
   ldh-str        =  *( utf8-rtext / "-" ) utf8-rtext
   utf8-rtext     =  ALPHA / DIGIT / UTF8-xtra-char

2.3. NAI Length Considerations

Devices handling NAIs MUST support an NAI length of at least 72 octets. Devices SHOULD support an NAI length of 253 octets. However, the following implementation issues should be considered:

  • NAI octet length constraints may impose a more severe constraint on the number of UTF-8 characters.
  • NAIs are often transported in the User-Name attribute of the Remote Authentication Dial-In User Service (RADIUS) protocol. Unfortunately, [RFC2865], Section 5.1 states that "the ability to handle at least 63 octets is recommended." As a result, it may not be possible to transfer NAIs beyond 63 octets through all devices. In addition, since only a single User-Name attribute may be included in a RADIUS message and the maximum attribute length is 253 octets, RADIUS is unable to support NAI lengths beyond 253 octets.
  • NAIs can also be transported in the User-Name attribute of Diameter [RFC6733], which supports content lengths up to 2^24 - 9 octets. As a result, NAIs processed only by Diameter nodes can be very long. However, an NAI transported over Diameter may eventually be translated to RADIUS, in which case the above limitations will apply.
  • NAIs may be transported in other protocols. Each protocol can have its own limitations on maximum NAI length.

The above criteria should permit the widest use and widest possible interoperability of the NAI.

2.4. Support for Username Privacy

Interpretation of the username part of the NAI depends on the realm in question. Therefore, the utf8-username portion SHOULD be treated as opaque data when processed by nodes that are not a part of the home domain for that realm.

That is, the only domain that is capable of interpreting the meaning of the utf8-username portion of the NAI is the home domain. Any third-party domains cannot form any conclusions about the utf8-username and cannot decode it into subfields. For example, it may be used as "firstname.lastname", or it may be entirely digits, or it may be a random hex identifier. There is simply no way (and no reason) for any other domain to interpret the utf8-username field as having any meaning whatsoever.

In some situations, NAIs are used together with a separate authentication method that can transfer the username part in a more secure manner to increase privacy. In this case, NAIs MAY be provided in an abbreviated form by omitting the username part. Omitting the username part is RECOMMENDED over using a fixed username part, such as "anonymous", since including a fixed username part is ambiguous as to whether or not the NAI refers to a single user. However, current practice is to use the username "anonymous" instead of omitting the username part. This behavior is also permitted.

The most common use case of omitting or obfuscating the username part is with TLS-based EAP methods such as Tunneled Transport Layer Security (TTLS) [RFC5281]. Those methods allow for an "outer" identifier, which is typically an anonymous "@realm". This outer identifier allows the authentication request to be routed from a visited domain to a home domain. At the same time, the username part is kept confidential from the visited network. The protocol provides for an "inner" authentication exchange, in which a full identifier is used to authenticate a user.

That scenario offers the best of both worlds. An anonymous NAI can be used to route authentication to the home domain, and the home domain has sufficient information to identify and authenticate users.

However, some protocols do not support authentication methods that allow for "inner" and "outer" exchanges. Those protocols are limited to using an identifier that is publicly visible. It is therefore RECOMMENDED that such protocols use ephemeral identifiers. We recognize that this practice is not currently used and will likely be difficult to implement.

Similar to the anonymous user, there may be situations where portions of the realm are sensitive. For those situations, it is RECOMMENDED that the sensitive portion of the realm also be omitted (e.g., to use "@example.com" instead of "@sensitive.example.com", or "anonymous@sensitive.example.com"). The home domain is authoritative for users in all subdomains and can (if necessary) route the authentication request to the appropriate subsystem within the home domain.

For roaming purposes, it is typically necessary to locate the appropriate backend authentication server for the given NAI before the authentication conversation can proceed. As a result, authentication routing is impossible unless the realm portion is available and is in a well-known format.

2.5. International Character Sets

This specification allows both international usernames and realms. International usernames are based on the use of Unicode characters, encoded as UTF-8. Internationalization of the username portion of the NAI is based on the "Internationalized Email Headers" [RFC6532] extensions to the "local-part" portion of email addresses [RFC5322].

In order to ensure a canonical representation, characters of the realm portion in an NAI MUST match the ABNF in this specification as well as the requirements specified in [RFC5891]. In practice, these requirements consist of the following item:

  • Realms MUST be of the form that can be registered as a Fully Qualified Domain Name (FQDN) within the DNS.

This list is significantly shorter and simpler than the list in Section 2.4 of [RFC4282]. The form suggested in [RFC4282] depended on intermediate nodes performing canonicalizations based on insufficient information, which meant that the form was not canonical.

Specifying the realm requirement as above means that the requirements depend on specifications that are referenced here, rather than copied here. This allows the realm definition to be updated when the referenced documents change, without requiring a revision of this specification.

One caveat on the above recommendation is the issues noted in [RFC6912]. That document notes that there are additional restrictions around DNS registration that forbid some code points from being valid in a DNS U-label. These restrictions cannot be expressed algorithmically.

For this specification, that caveat means the following: Realms not matching the above ABNF are not valid NAIs. However, some realms that do match the ABNF are still invalid NAIs. That is, matching the ABNF is a necessary, but not sufficient, requirement for an NAI.

In general, the above requirement means following the requirements specified in [RFC5891].

2.6. The Normalization Process

Conversion to Unicode as well as normalization SHOULD be performed by edge systems (e.g., laptops, desktops, smart phones, etc.) that take "local" text as input. These edge systems are best suited to determine the user's intent and can best convert from "local" text to a normalized form.

Other AAA systems such as proxies do not have access to locale and character set information that is available to edge systems. Therefore, they may not always be able to convert local input to Unicode.

That is, all processing of NAIs from "local" character sets and locales to UTF-8 SHOULD be performed by edge systems, prior to the NAIs entering the AAA system. Inside of a AAA system, NAIs are sent over the wire in their canonical form, and this canonical form is used for all NAI and/or realm comparisons.

Copying of localized text into fields that can subsequently be placed into the RADIUS User-Name attribute is problematic. This practice can result in a AAA proxy encountering non-UTF-8 characters within what it expects to be an NAI. An example of this requirement is Section 2.1 of [RFC3579], which states:

the NAS MUST copy the contents of the Type-Data field of the EAP-Response/Identity received from the peer into the User-Name attribute

As a result, AAA proxies expect the contents of the EAP-Response/Identity sent by an EAP supplicant to consist of UTF-8 characters, not localized text. Using localized text in AAA username or identity fields means that realm routing becomes difficult or impossible.

In contrast to Section 2.4 of [RFC4282], AAA systems are now expected to perform NAI comparisons, matching, and AAA routing based on the NAI as it is received. This specification provides a canonical representation, ensures that intermediate AAA systems such as proxies are not required to perform translations, and can be expected to work through AAA systems that are unaware of international character sets.

In an ideal world, the following requirements would be widely implemented:

  • Edge systems using "localized" text SHOULD normalize the NAI prior to it being used as an identifier in an authentication protocol.
  • AAA systems SHOULD NOT normalize the NAI, as they may not have sufficient information to perform the normalization.

There are issues with this approach, however.

2.6.1. Issues with the Normalization Process

The requirements in the preceding section are not implemented today. For example, most EAP implementations use a user identifier that is passed to them from some other local system. This identifier is treated as an opaque blob and is placed as is into the EAP Identity field. Any subsequent system that receives that identifier is assumed to be able to understand and process it.

This opaque blob unfortunately can contain localized text, which means that the AAA systems have to process that text.

These limitations have the following theoretical and practical implications:

  • Edge systems used today generally do not normalize the NAI.
  • Therefore, AAA systems SHOULD attempt to normalize the NAI.

The suggestions above contradict the suggestions in the previous section. This is the reality of imperfect protocols.

Where the user identifier can be normalized, or determined to be in normal form, the normal form MUST be used as the NAI. In all other circumstances, the user identifier MUST NOT be treated as an NAI. That data is still, however, a user identifier. AAA systems MUST NOT fail authentication simply because the user identifier is not an NAI.

That is, when the realm portion of the NAI is not recognized by a AAA server, it SHOULD try to normalize the NAI into NFC form. That normalized form can then be used to see if the realm matches a known realm. If no match is found, the original form of the NAI SHOULD be used in all subsequent processing.

The AAA server may also convert realms to Punycode and perform all realm comparisons on the resulting Punycode strings. This conversion follows the recommendations above but may have different operational effects and failure modes.

2.7. Use in Other Protocols

As noted earlier, the NAI format can be used in other, non-AAA protocols. It is RECOMMENDED that the definition given here be used unchanged. Using other definitions for user identifiers may hinder interoperability, along with the user's ability to authenticate successfully. It is RECOMMENDED that protocols requiring the use of a user identifier use the NAI format.

This document cannot require other protocols to use the NAI format for user identifiers. Their needs are unknown and, at this time, unknowable. This document suggests that interoperability and inter-domain authentication are useful and should be encouraged.

Where a protocol is 8-bit clean, it can likely transport the NAI as is, without further modification.

Where a protocol is not 8-bit clean, it cannot transport the NAI as is. Instead, this document presumes that a protocol-specific transport layer takes care of encoding the NAI on input to the protocol and decoding it when the NAI exits the protocol. The encoded or escaped version of the NAI is not a valid NAI and MUST NOT be presented to the AAA system.

For example, HTTP carries user identifiers but escapes the '.' character as "%2E" (among others). When HTTP is used to transport the NAI "fred@example.com", the data as transported will be in the form "fred@example%2Ecom". That data exists only within HTTP and has no relevance to any AAA system.

Any comparison, validation, or use of the NAI MUST be done on its unescaped (i.e., utf8-clean) form.

2.8. Using the NAI Format for Other Identifiers

As discussed in Section 1, above, it is RECOMMENDED that the NAI format be used as the standard format for user identifiers. This section discusses that use in more detail.

It is often useful to create new identifiers for use in specific contexts. These identifiers may have a number of different properties, most of which are unimportant to this document. The goal of this document is to create identifiers that are to be in a well-known format and that will have namespaces. The NAI format fits these requirements.

One example of such use is the "private user identity", which is an identifier defined by the 3rd Generation Partnership Project (3GPP). That identifier is used to uniquely identify the user to the network. The identifier is used for authorization, authentication, accounting, administration, etc. The "private user identity" is globally unique and is defined by the home network operator. The format of the identifier is explicitly the NAI, as stated by Section 13.3 of [3GPP]:

The private user identity shall take the form of an NAI, and shall have the form username@realm as specified in clause 2.1 of IETF RFC 4282

For 3GPP, the "username" portion is a unique identifier that is derived from device-specific information. The "realm" portion is composed of information about the home network, followed by the base string "3gppnetwork.org" (e.g., 234150999999999@ims.mnc015.mcc234.3gppnetwork.org).

This format as defined by 3GPP ensures that the identifier is globally unique, as it is based on the "3gppnetwork.org" domain. It ensures that the "realm" portion is specific to a particular home network (or organization), via the "ims.mnc015.mcc234" prefix to the realm. Finally, it ensures that the "username" portion follows a well-known format.

This document suggests that the NAI format be used for all new specifications and/or protocols where a user identifier is required. Where the username portions need to be created with subfields, a well-known and documented method, as has been done with 3GPP, is preferred to ad hoc methods.

3. Routing inside of AAA Systems

Many AAA systems use the "utf8-realm" portion of the NAI to route requests within a AAA proxy network. The semantics of this operation involves a logical AAA routing table, where the "utf8-realm" portion acts as a key, and the values stored in the table are one or more "next hop" AAA servers.

Intermediate nodes MUST use the "utf8-realm" portion of the NAI without modification to perform this lookup. As noted earlier, intermediate nodes may not have access to the same locale information as the system that injected the NAI into the AAA routing systems. Therefore, almost all "case insensitive" comparisons can be wrong. Where the "utf8-realm" is entirely ASCII, current AAA systems sometimes perform case-insensitive matching on realms. This method MAY be continued, as it has been shown to work in practice.

Many existing non-AAA systems have user identifiers that are similar in format to the NAI but that are not compliant with this specification. For example, they may use non-NFC form, or they may have multiple "@" characters in the user identifier. Intermediate nodes SHOULD normalize non-NFC identifiers to NFC, prior to looking up the "utf8-realm" in the logical routing table. Intermediate nodes MUST NOT modify the identifiers that they forward. The data as entered by the user is inviolate.

The "utf8-realm" provisioned in the logical AAA routing table SHOULD be provisioned to the proxy prior to it receiving any AAA traffic. The "utf8-realm" SHOULD be supplied by the "next hop" or "home" system that also supplies the routing information necessary for packets to reach the next hop.

This "next hop" information may be any of, or all of, the following information: IP address, port, RADIUS shared secret, TLS certificate, DNS host name, or instruction to use dynamic DNS discovery (i.e., look up a record in the "utf8-realm" domain). This list is not exhaustive and may be extended by future specifications.

It is RECOMMENDED to use the entirety of the "utf8-realm" for the routing decisions. However, AAA systems MAY use a portion of the "utf8-realm" portion, so long as that portion is a valid "utf8-realm" and is handled as above. For example, routing "fred@example.com" to a "com" destination is forbidden, because "com" is not a valid "utf8-realm". However, routing "fred@sales.example.com" to the "example.com" destination is permissible.

Another reason to forbid the use of a single label (e.g., "fred@sales") is that many non-AAA systems treat a single label as being a local identifier within their realm. That is, a user logging in as "fred@sales" to a domain "example.com" would be treated as if the NAI was instead "fred@sales.example.com". Permitting the use of a single label would mean changing the interpretation and meaning of a single label, which cannot be done.

3.1. Compatibility with Email Usernames

As proposed in this document, the Network Access Identifier is of the form "user@realm". Please note that while the user portion of the NAI is based on the "Internet Message Format" [RFC5322] "local-part" portion of an email address as extended by "Internationalized Email Headers" [RFC6532], it has been modified for the purposes of Section 2.2. It does not permit quoted text along with "folding" or "non-folding" whitespace that is commonly used in email addresses. As such, the NAI is not necessarily equivalent to usernames used in email.

However, it is a common practice to use email addresses as user identifiers in AAA systems. The ABNF in Section 2.2 is defined to be close to the "addr-spec" portion of [RFC5322] as extended by [RFC6532], while still being compatible with [RFC4282].

In contrast to Section 2.5 of [RFC4282], this document states that the internationalization requirements for NAIs and email addresses are substantially similar. The NAI and email identifiers may be the same, and both need to be entered by the user and/or the operator supplying network access to that user. There is therefore good reason for the internationalization requirements to be similar.

3.2. Compatibility with DNS

The "utf8-realm" portion of the NAI is intended to be compatible with Internationalized Domain Names (IDNs) [RFC5890]. As defined above, the "utf8-realm" portion as transported within an 8-bit clean protocol such as RADIUS and EAP can contain any valid UTF-8 character. There is therefore no reason for a NAS to convert the "utf8-realm" portion of an NAI into Punycode encoding form [RFC3492] prior to placing the NAI into a RADIUS User-Name attribute.

The NAI does not make a distinction between A-labels and U-labels, as those are terms specific to DNS. It is instead an IDNA-valid label, as per the first item in Section 2.3.2.1 of [RFC5890]. As noted in that section, the term "IDNA-valid label" encompasses both "A-label" and "U-label".

When the realm portion of the NAI is used as the basis for name resolution, it may be necessary to convert internationalized realm names to Punycode [RFC3492] encoding form as described in [RFC5891]. As noted in Section 2 of [RFC6055], resolver Application Programming Interfaces (APIs) are not necessarily DNS specific, so conversion to Punycode needs to be done carefully:

Applications that convert an IDN to A-label form before calling (for example) getaddrinfo() will result in name resolution failures if the Punycode name is directly used in such protocols. Having libraries or protocols to convert from A-labels to the encoding scheme defined by the protocol (e.g., UTF-8) would require changes to APIs and/or servers, which Internationalized Domain Names for Applications (IDNA) was intended to avoid.

As a result, applications SHOULD NOT assume that non-ASCII names are resolvable using the public DNS and blindly convert them to A-labels without knowledge of what protocol will be selected by the name resolution library.

3.3. Realm Construction

The home realm usually appears in the "utf8-realm" portion of the NAI, but in some cases a different realm can be used. This may be useful, for instance, when the home realm is reachable only via intermediate proxies.

Such usage may prevent interoperability unless the parties involved have a mutual agreement that the usage is allowed. In particular, NAIs MUST NOT use a different realm than the home realm unless the sender has explicit knowledge that (a) the specified other realm is available and (b) the other realm supports such usage. The sender may determine the fulfillment of these conditions through a database, dynamic discovery, or other means not specified here. Note that the first condition is affected by roaming, as the availability of the other realm may depend on the user's location or the desired application.

The use of the home realm MUST be the default unless otherwise configured.

3.3.1. Historical Practices

Some AAA systems have historically used NAI modifications with multiple "prefix" and "suffix" decorations to perform explicit routing through multiple proxies inside of a AAA network.

In RADIUS-based environments, the use of decorated NAI is NOT RECOMMENDED for the following reasons:

  • Using explicit routing paths is fragile and is unresponsive to changes in the network due to servers going up or down or to changing business relationships.
  • There is no RADIUS routing protocol, meaning that routing paths have to be communicated "out of band" to all intermediate AAA nodes, and also to all edge systems (e.g., supplicants) expecting to obtain network access.
  • Using explicit routing paths requires thousands, if not millions, of edge systems to be updated with new path information when a AAA routing path changes. This adds huge expense for updates that would be better done at only a few AAA systems in the network.
  • Manual updates to RADIUS paths are expensive, time-consuming, and prone to error.
  • Creating compatible formats for the NAI is difficult when locally defined "prefixes" and "suffixes" conflict with similar practices elsewhere in the network. These conflicts mean that connecting two networks may be impossible in some cases, as there is no way for packets to be routed properly in a way that meets all requirements at all intermediate proxies.
  • Leveraging the DNS name system for realm names establishes a globally unique namespace for realms.

In summary, network practices and capabilities have changed significantly since NAIs were first overloaded to define AAA routes through a network. While manually managed explicit path routing was once useful, the time has come for better methods to be used.

Notwithstanding the above recommendations, the above practice is widely used for Diameter routing [RFC5729]. The routes described there are managed automatically, for both credential provisioning and routing updates. Those routes also exist within a particular framework (typically 3G), where membership is controlled and system behavior is standardized. There are no known issues with using explicit routing in such an environment.

However, if decorated identifiers are used, such as: 

      homerealm.example.org!user@otherrealm.example.net

then the part before the (non-escaped) '!' MUST be a "utf8-realm" as defined in the ABNF in Section 2.2. When receiving such an identifier, the "otherrealm.example.net" system MUST convert the identifier to "user@homerealm.example.org" before forwarding the request. The forwarding system MUST then apply normal AAA routing for the transaction, based on the updated identifier.

3.4. Examples

Examples of valid Network Access Identifiers include the following: 

           bob
           joe@example.com
           fred@foo-9.example.com
           jack@3rd.depts.example.com
           fred.smith@example.com
           fred_smith@example.com
           fred$@example.com
           fred=?#$&*+-/^smith@example.com
           nancy@eng.example.net
           eng.example.net!nancy@example.net
           eng%nancy@example.net
           @privatecorp.example.net
           \(user\)@example.net

An additional valid NAI is the following -- shown here as a hex string, as this document can only contain ASCII characters:

626f 6240 ceb4 cebf ceba ceb9 cebc ceae 2e63 6f6d

Examples of invalid Network Access Identifiers include the following: 

           fred@example
           fred@example_9.com
           fred@example.net@example.net
           fred.@example.net
           eng:nancy@example.net
           eng;nancy@example.net
           (user)@example.net
           <nancy>@example.net

One example given in [RFC4282] is still permitted by the ABNF, but it is NOT RECOMMENDED because of the use of the Punycode [RFC3492] encoding form for what is now a valid UTF-8 string: 

           alice@xn--tmonesimerkki-bfbb.example.net

4. Security Considerations

Since an NAI reveals the home affiliation of a user, it may assist an attacker in further probing the username space. Typically, this problem is of most concern in protocols that transmit the username in clear-text across the Internet, such as in RADIUS [RFC2865] [RFC2866]. In order to prevent snooping of the username, protocols may use confidentiality services provided by protocols transporting them, such as RADIUS protected by IPsec [RFC3579] or Diameter protected by TLS [RFC6733].

This specification adds the possibility of hiding the username part in the NAI, by omitting it. As discussed in Section 2.4, this is possible only when NAIs are used together with a separate authentication method that can transfer the username in a secure manner. In some cases, application-specific privacy mechanisms have also been used with NAIs. For instance, some EAP methods apply method-specific pseudonyms in the username part of the NAI [RFC3748]. While neither of these approaches can protect the realm part, their advantage over transport protection is that the privacy of the username is protected, even through intermediate nodes such as NASes.

4.1. Correlation of Identities over Time and Protocols

The recommendations in Sections 2.7 and 2.8 for using the NAI in other protocols have implications for privacy. Any attacker who is capable of observing traffic containing the NAI can track the user and can correlate his activity across time and across multiple protocols. The authentication credentials therefore SHOULD be transported over channels that permit private communications, or multiple identifiers SHOULD be used, so that user tracking is impossible.

It is RECOMMENDED that user privacy be enhanced by configuring multiple identifiers for one user. These identifiers can be changed over time, in order to make user tracking more difficult for a malicious observer. However, provisioning and management of the identifiers may be difficult to do in practice -- a likely reason why multiple identifiers are rarely used today.

4.2. Multiple Identifiers

Section 1.3 states that multiple identifier formats allow attackers to make contradictory claims without being detected. This statement deserves further discussion.

Section 2.4 discussed "inner" and "outer" identifiers in the context of TTLS [RFC5281]. A close reading of that specification shows there is no requirement that the inner and outer identifiers be in any way related. That is, it is perfectly valid to use "@example.com" for an outer identifier and "user@example.org" as an inner identifier. The authentication request will then be routed to "example.com", which will likely be unable to authenticate "user@example.org".

Even worse, a misconfiguration of "example.com" means that it may in turn proxy the inner authentication request to the "example.org" domain. Such cross-domain authentication is highly problematic, and there are few good reasons to allow it.

It is therefore RECOMMENDED that systems that permit anonymous "outer" identifiers require that the "inner" domain be the same as, or a subdomain of, the "outer" domain. An authentication request using disparate realms is a security violation, and the request SHOULD be rejected.

The situation gets worse when multiple protocols are involved. The TTLS protocol permits Microsoft CHAP (MS-CHAP) [RFC2433] to be carried inside of the TLS tunnel. MS-CHAP defines its own identifier, which is encapsulated inside of the MS-CHAP exchange. That identifier is not required to be any particular format, is not required to be in UTF-8, and, in practice, can be one of many unknown character sets. There is no way in practice to determine which character set was used for that identifier.

The result is that the "outer" EAP Identity carried by TTLS is likely to not even share the same character set as the "inner" identifier used by MS-CHAP. The two identifiers are entirely independent and fundamentally incomparable.

Such a protocol design is NOT RECOMMENDED.

5. Administration of Names

In order to avoid creating any new administrative procedures, administration of the NAI realm namespace piggybacks on the administration of the DNS namespace.

NAI realm names are required to be unique, and the rights to use a given NAI realm for roaming purposes are obtained coincident with acquiring the rights to use a particular Fully Qualified Domain Name (FQDN). Those wishing to use an NAI realm name should first acquire the rights to use the corresponding FQDN. Administrators MUST NOT publicly use an NAI realm without first owning the corresponding FQDN. Private use of unowned NAI realms within an administrative domain is allowed, though it is RECOMMENDED that example names be used, such as "example.com".

Note that the use of an FQDN as the realm name does not require use of the DNS for location of the authentication server. While Diameter [RFC6733] supports the use of DNS for location of authentication servers, existing RADIUS implementations typically use proxy configuration files in order to locate authentication servers within a domain and perform authentication routing. The implementations described in [RFC2194] did not use DNS for location of the authentication server within a domain. Similarly, existing implementations have not found a need for dynamic routing protocols or propagation of global routing information. Note also that there is no requirement that the NAI represent a valid email address.

6. References

6.1. Normative References

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119, March 1997,
              <http://www.rfc-editor.org/info/rfc2119>.
   
   [RFC3629]  Yergeau, F., "UTF-8, a transformation format of
              ISO 10646", STD 63, RFC 3629, November 2003,
              <http://www.rfc-editor.org/info/rfc3629>.
   
   [RFC5198]  Klensin, J. and M. Padlipsky, "Unicode Format for Network
              Interchange", RFC 5198, March 2008,
              <http://www.rfc-editor.org/info/rfc5198>.
   
   [RFC5234]  Crocker, D., Ed., and P. Overell, "Augmented BNF for
              Syntax Specifications: ABNF", STD 68, RFC 5234,
              January 2008, <http://www.rfc-editor.org/info/rfc5234>.
   
   [RFC5890]  Klensin, J., "Internationalized Domain Names for
              Applications (IDNA): Definitions and Document Framework",
              RFC 5890, August 2010,
              <http://www.rfc-editor.org/info/rfc5890>.
   
   [RFC5891]  Klensin, J., "Internationalized Domain Names in
              Applications (IDNA): Protocol", RFC 5891, August 2010,
              <http://www.rfc-editor.org/info/rfc5891>.
   
   [RFC6365]  Hoffman, P. and J. Klensin, "Terminology Used in
              Internationalization in the IETF", BCP 166, RFC 6365,
              September 2011, <http://www.rfc-editor.org/info/rfc6365>.

6.2. Informative References

   [RFC2194]  Aboba, B., Lu, J., Alsop, J., Ding, J., and W. Wang,
              "Review of Roaming Implementations", RFC 2194,
              September 1997, <http://www.rfc-editor.org/info/rfc2194>.
   
   [RFC2341]  Valencia, A., Littlewood, M., and T. Kolar, "Cisco
              Layer Two Forwarding (Protocol) "L2F"", RFC 2341,
              May 1998, <http://www.rfc-editor.org/info/rfc2341>.
   
   [RFC2433]  Zorn, G. and S. Cobb, "Microsoft PPP CHAP Extensions",
              RFC 2433, October 1998,
              <http://www.rfc-editor.org/info/rfc2433>.
   
   [RFC2637]  Hamzeh, K., Pall, G., Verthein, W., Taarud, J., Little,
              W., and G. Zorn, "Point-to-Point Tunneling Protocol
              (PPTP)", RFC 2637, July 1999,
              <http://www.rfc-editor.org/info/rfc2637>.
   
   [RFC2661]  Townsley, W., Valencia, A., Rubens, A., Pall, G., Zorn,
              G., and B. Palter, "Layer Two Tunneling Protocol "L2TP"",
              RFC 2661, August 1999,
              <http://www.rfc-editor.org/info/rfc2661>.
   
   [RFC2865]  Rigney, C., Willens, S., Rubens, A., and W. Simpson,
              "Remote Authentication Dial In User Service (RADIUS)",
              RFC 2865, June 2000,
              <http://www.rfc-editor.org/info/rfc2865>.
   
   [RFC2866]  Rigney, C., "RADIUS Accounting", RFC 2866, June 2000,
              <http://www.rfc-editor.org/info/rfc2866>.
   
   [RFC3492]  Costello, A., "Punycode: A Bootstring encoding of Unicode
              for Internationalized Domain Names in Applications
              (IDNA)", RFC 3492, March 2003,
              <http://www.rfc-editor.org/info/rfc3492>.
   
   [RFC3579]  Aboba, B. and P. Calhoun, "RADIUS (Remote Authentication
              Dial In User Service) Support For Extensible
              Authentication Protocol (EAP)", RFC 3579, September 2003,
              <http://www.rfc-editor.org/info/rfc3579>.
   
   [RFC3748]  Aboba, B., Blunk, L., Vollbrecht, J., Carlson, J., and H.
              Levkowetz, Ed., "Extensible Authentication Protocol
              (EAP)", RFC 3748, June 2004,
              <http://www.rfc-editor.org/info/rfc3748>.
   
   [RFC4282]  Aboba, B., Beadles, M., Arkko, J., and P. Eronen, "The
              Network Access Identifier", RFC 4282, December 2005,
              <http://www.rfc-editor.org/info/rfc4282>.
   
   [RFC4301]  Kent, S. and K. Seo, "Security Architecture for the
              Internet Protocol", RFC 4301, December 2005,
              <http://www.rfc-editor.org/info/rfc4301>.
   
   [RFC5281]  Funk, P. and S. Blake-Wilson, "Extensible Authentication
              Protocol Tunneled Transport Layer Security Authenticated
              Protocol Version 0 (EAP-TTLSv0)", RFC 5281, August 2008,
              <http://www.rfc-editor.org/info/rfc5281>.
   
   [RFC5322]  Resnick, P., Ed., "Internet Message Format", RFC 5322,
              October 2008, <http://www.rfc-editor.org/info/rfc5322>.
   
   [RFC5335]  Yang, A., Ed., "Internationalized Email Headers",
              RFC 5335, September 2008,
              <http://www.rfc-editor.org/info/rfc5335>.
   
   [RFC5729]  Korhonen, J., Ed., Jones, M., Morand, L., and T. Tsou,
              "Clarifications on the Routing of Diameter Requests Based
              on the Username and the Realm", RFC 5729, December 2009,
              <http://www.rfc-editor.org/info/rfc5729>.
   
   [RFC6055]  Thaler, D., Klensin, J., and S. Cheshire, "IAB Thoughts on
              Encodings for Internationalized Domain Names", RFC 6055,
              February 2011, <http://www.rfc-editor.org/info/rfc6055>.
   
   [RFC6532]  Yang, A., Steele, S., and N. Freed, "Internationalized
              Email Headers", RFC 6532, February 2012,
              <http://www.rfc-editor.org/info/rfc6532>.
   
   [RFC6733]  Fajardo, V., Ed., Arkko, J., Loughney, J., and G. Zorn,
              Ed., "Diameter Base Protocol", RFC 6733, October 2012,
              <http://www.rfc-editor.org/info/rfc6733>.
   
   [RFC6912]  Sullivan, A., Thaler, D., Klensin, J., and O. Kolkman,
              "Principles for Unicode Code Point Inclusion in Labels in
              the DNS", RFC 6912, April 2013,
              <http://www.rfc-editor.org/info/rfc6912>.
   
   [EDUROAM]  "eduroam (EDUcation ROAMing)", <http://eduroam.org>.
   
   [3GPP]     3GPP, "Numbering, addressing and Identification", 3GPP TS
              23.003, Release 12, July 2014,
              <ftp://ftp.3gpp.org/Specs/archive/23_series/23.003/>.

Appendix A. Changes from RFC 4282

This document contains the following updates with respect to the previous NAI definition in RFC 4282 [RFC4282]:

  • The formal syntax in Section 2.1 has been updated to forbid non-UTF-8 characters (e.g., characters with the "high bit" set).
  • The formal syntax in Section 2.1 of [RFC4282] has been updated to allow UTF-8 in the "realm" portion of the NAI.
  • The formal syntax in Section 2.1 of [RFC4282] applied to the NAI after it was "internationalized" via the ToAscii function. The contents of the NAI before it was "internationalized" were left indeterminate. This document updates the formal syntax to define an internationalized form of the NAI and forbids the use of the ToAscii function for NAI "internationalization".
  • The grammar for the user and realm portion is based on a combination of the "nai" defined in Section 2.1 of [RFC4282] and the "utf8-addr-spec" defined in Section 4.4 of [RFC5335].
  • All use of the ToAscii function has been moved to normal requirements on DNS implementations when realms are used as the basis for DNS lookups. This involves no changes to the existing DNS infrastructure.
  • The discussions on internationalized character sets in Section 2.4 of [RFC4282] have been updated. The suggestion to use the ToAscii function for realm comparisons has been removed. No AAA system has implemented these suggestions, so this change should have no operational impact.
  • The "Routing inside of AAA Systems" section is new in this document. The concept of a "local AAA routing table" is also new, although it accurately describes the functionality of widespread implementations.
  • The "Compatibility with EMail Usernames" and "Compatibility with DNS" sections have been revised and updated. The Punycode transformation is suggested to be used only when a realm name is used for DNS lookups, and even then the function is only used by a resolving API on the local system, and even then it is recommended that only the home network perform this conversion.
  • The "Realm Construction" section has been updated to note that editing of the NAI is NOT RECOMMENDED.
  • The "Examples" section has been updated to remove the instance of the IDN being converted to ASCII. This behavior is now forbidden.

Acknowledgments

The initial text for this document was [RFC4282], which was then heavily edited. The original authors of [RFC4282] were Bernard Aboba, Mark A. Beadles, Jari Arkko, and Pasi Eronen.

Author's Address

Alan DeKok
The FreeRADIUS Server Project

EMail: 

          aland@freeradius.org