Verisign Labs

12061 Bluemont Way Reston VA 20190 United States shollenbeck@verisign.com http://www.verisignlabs.com/

China Internet Network Information Center

4 South 4th Street, Zhongguancun, Haidian District China 100190 Beijing +86 10 5881 3147 nkong@cnnic.cn

Applications RDAP Security The Registration Data Access Protocol (RDAP) provides "RESTful" web services to retrieve registration metadata from Domain Name and Regional Internet Registries. This document describes information security services, including access control, authentication, authorization, availability, data confidentiality, and data integrity for RDAP.

The Registration Data Access Protocol (RDAP) is specified in multiple documents, including "Registration Data Access Protocol (RDAP) Query Format" , "JSON Responses for the Registration Data Access Protocol (RDAP)" , and "HTTP Usage in the Registration Data Access Protocol (RDAP)" . One goal of RDAP is to provide security services that do not exist in the WHOIS protocol, including access control, authentication, authorization, availability, data confidentiality, and data integrity. This document describes how each of these services is achieved by RDAP using features that are available in other protocol layers. Additional or alternative mechanisms can be added in the future. Where applicable, informative references to requirements for a WHOIS replacement service are noted.

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

DNR: Domain Name Registry HTTP: Hypertext Transfer Protocol JSON: JavaScript Object Notation RDAP: Registration Data Access Protocol RIR: Regional Internet Registry TLS: Transport Layer Security

RDAP itself does not include native security services. Instead, RDAP relies on features that are available in other protocol layers to provide needed security services, including access control, authentication, authorization, availability, data confidentiality, and data integrity. A description of each of these security services can be found in "Internet Security Glossary, Version 2" . No requirements have been identified for other security services.

WHOIS does not include specific features to control access to registration information. As described in the following sections, RDAP includes features to identify, authenticate, and authorize clients, allowing server operators to control access to information based on a client's identity and associated authorizations. Information returned to a client can be clearly marked with a status value (see Section 10.2.2 of ) that identifies the access granted to the client.

This section describes security authentication mechanisms and the need for authorization policies to include them. It describes requirements for the implementations of clients and servers but does not dictate the policies of server operators. For example, a server operator with no policy regarding differentiated or tiered access to data will have no authorization mechanisms and will have no need for any type of authentication. A server operator with policies on differentiated access will have to construct an authorization scheme and will need to follow the specified authentication requirements. WHOIS does not provide features to identify and authenticate clients. As noted in Section 3.1.4.2 of "Cross Registry Internet Service Protocol (CRISP) Requirements" , there is utility in allowing server operators to offer "varying degrees of access depending on policy and need." Clients have to be identified and authenticated to provide that utility. RDAP's authentication framework needs to accommodate anonymous access as well as verification of identities using a range of authentication methods and credential services. To that end, RDAP clients and servers MUST implement the authentication framework specified in "Hypertext Transfer Protocol (HTTP/1.1): Authentication" . The "basic" scheme can be used to send a client's user name and password to a server in plaintext, base64-encoded form. The "digest" scheme can be used to authenticate a client without exposing the client's plaintext password. If the "basic" scheme is used, HTTP over TLS MUST be used to protect the client's credentials from disclosure while in transit (see ). Servers MUST support either Basic or Digest authentication; they are not required to support both. Clients MUST support both to interoperate with servers that support one or the other. Servers may provide a login page that triggers HTTP authentication. Clients should continue sending the HTTP authentication header once they receive an initial 401 (Unauthorized) response from the HTTP server as long as the scheme portion of the URL doesn't change. The Transport Layer Security protocol includes an optional feature to identify and authenticate clients who possess and present a valid X.509 digital certificate . Support for this feature is OPTIONAL. RDAP does not impose any unique server authentication requirements. The server authentication provided by TLS fully addresses the needs of RDAP. In general, transports for RDAP must either provide a TLS&nbhy;protected transport (e.g., HTTPS) or a mechanism that provides an equivalent level of server authentication. Work on HTTP authentication methods continues. RDAP is designed to be agile enough to support additional methods as they are defined.

The traditional client-server authentication model requires clients to maintain distinct credentials for every RDAP server. This situation can become unwieldy as the number of RDAP servers increases. Federated authentication mechanisms allow clients to use one credential to access multiple RDAP servers and reduce client credential management complexity. RDAP MAY include a federated authentication mechanism that permits a client to access multiple RDAP servers in the same federation with one credential. Federated authentication mechanisms used by RDAP MUST be fully supported by HTTP. OAuth, OpenID, Security Assertion Markup Language (SAML), and mechanisms based on Certification Authority (CA) are all possible approaches to provide federated authentication. At the time of this document's publication, negotiation or advertisement of federated authentication services is still an undefined mechanism by the noted federated authentication protocols. Developing this mechanism is beyond the scope of this document. The OAuth authorization framework describes a method for users to access protected web resources without having to hand out their credentials. Instead, clients are issued access tokens by authorization servers with the permission of the resource owners. Using OAuth, multiple RDAP servers can form a federation, and the clients can access any server in the same federation by providing one credential registered in any server in that federation. The OAuth authorization framework is designed for use with HTTP and thus can be used with RDAP. OpenID is a decentralized single sign-on authentication system that allows users to log in at multiple web sites with one ID instead of having to create multiple unique accounts. An end user can freely choose which OpenID provider to use and can preserve their Identifier if they switch OpenID providers. Note that OAuth and OpenID do not consistently require data confidentiality services to protect interactions between providers and consumers. HTTP over TLS can be used as needed to provide protection against man-in-the-middle attacks. SAML 2.0 is an XML-based protocol that can be used to implement web-based authentication and authorization services, including single sign on. It uses security tokens containing assertions to exchange information about an end user between an identity provider and a service provider. The Transport Layer Security protocol describes the specification of a client certificate in Section 7.4.6 of . Clients who possess and present a valid X.509 digital certificate, issued by a CA, could be identified and authenticated by a server who trusts the corresponding CA. A certificate authentication method can be used to achieve federated authentication in which multiple RDAP servers all trust the same CAs, and then any client with a certificate issued by a trusted CA can access any RDAP server in the federation. This certificate-based mechanism is supported by HTTPS and can be used with RDAP.

WHOIS does not provide services to grant different levels of access to clients based on a client's authenticated identity. As noted in Section 3.1.4.2 of "Cross Registry Internet Service Protocol (CRISP) Requirements" , there is utility in allowing server operators to offer "varying degrees of access depending on policy and need." Access control decisions can be made once a client's identity has been established and authenticated (see ). Server operators MAY offer varying degrees of access depending on policy and need in conjunction with the authentication methods described in . If such varying degrees of access are supported, an RDAP server MUST provide granular access controls (that is, per registration data object) in order to implement authorization policies. Some examples: Clients will be allowed access only to data for which they have a relationship. Unauthenticated or anonymous access status may not yield any contact information. Full access may be granted to a special group of authenticated clients. The type of access allowed by a server will most likely vary from one operator to the next. A description of the response privacy considerations associated with different levels of authorization can be found in Section 13 of .

An RDAP service has to be available to be useful. There are no RDAP-unique requirements to provide availability, but as a general security consideration, a service operator needs to be aware of the issues associated with denial of service. A thorough reading of "Internet Denial-of-Service Considerations" is advised. An RDAP service MAY use an HTTP throttling mechanism to limit the number of queries that a single client can send in a given period of time. If used, the server SHOULD return an HTTP 429 (Too Many Requests) response code as described in "Additional HTTP Status Codes" . A client that receives a 429 response SHOULD decrease its query rate and honor the Retry-After header field if one is present. Note that this is not a defense against denial&nbhy;of&nbhy;service attacks, since a malicious client could ignore the code and continue to send queries at a high rate. A server might use another response code if it did not wish to reveal to a client that rate limiting is the reason for the denial of a reply.

WHOIS does not provide the ability to protect data from inadvertent disclosure while in transit. RDAP uses HTTP over TLS to provide that protection by encrypting all traffic sent on the connection between client and server. HTTP over TLS MUST be used to protect all client-server exchanges unless operational constraints make it impossible to meet this requirement. It is also possible to encrypt discrete objects (such as command path segments and JSON-encoded response objects) at one endpoint, send them to the other endpoint via an unprotected transport protocol, and decrypt the object on receipt. Encryption algorithms as described in "Internet Security Glossary, Version 2" are commonly used to provide data confidentiality at the object level. There are no current requirements for object-level data confidentiality using encryption. Support for this feature could be added to RDAP in the future. As noted in , the HTTP "basic" authentication scheme can be used to authenticate a client. When this scheme is used, HTTP over TLS MUST be used to protect the client's credentials from disclosure while in transit. If the policy of the server operator requires encryption to protect client-server data exchanges (such as to protect non-public data that cannot be accessed without client identification and authentication), HTTP over TLS MUST be used to protect those exchanges. A description of privacy threats that can be addressed with confidentiality services can be found in . Section 10.2.2 of describes status values that can be used to describe operator actions used to protect response data from disclosure to unauthorized clients.

WHOIS does not provide the ability to protect data from modification while in transit. Web services such as RDAP commonly use HTTP over TLS to provide that protection by using a keyed Message Authentication Code (MAC) to detect modifications. It is also possible to sign discrete objects (such as command path segments and JSON-encoded response objects) at one endpoint, send them to the other endpoint via a transport protocol, and validate the signature of the object on receipt. Digital signature algorithms as described in "Internet Security Glossary, Version 2" are commonly used to provide data integrity at the object level. There are no current requirements for object-level data integrity using digital signatures. Support for this feature could be added to RDAP in the future. The most specific need for this service is to provide assurance that HTTP 30x redirection hints and response elements returned from the server are not modified while in transit. If the policy of the server operator requires message integrity for client-server data exchanges, HTTP over TLS MUST be used to protect those exchanges.

Registration data has historically included personal data about registrants. WHOIS services have historically made this information available to the public, creating a privacy risk by revealing the personal details of registrants. WHOIS services have not had the benefit of authentication or access control mechanisms to gate access to registration data. As a result of this, proxy and privacy services have arisen to shield the identities of registrants. The standardization of RDAP does not change or impact the data that operators may require to be collected from registrants, but it provides support for a number of mechanisms that may be used to mitigate privacy threats to registrants should operators choose to use them. RDAP includes mechanisms that can be used to authenticate clients, allowing servers to support tiered access based on local policy. This means that all registration data need no longer be public, and personal data or data that may be considered more sensitive can have its access restricted to specifically privileged clients. RDAP data structures allow servers to indicate via status values when data returned to clients has been made private, redacted, obscured, or registered by a proxy. "Private" means that the data is not designated for public consumption. "Redacted" means that some registration data fields are not being made available. "Obscured" means that data has been altered for the purposes of not readily revealing the actual registration information. One option that operators have available to them to reduce privacy risks to registrants is to adopt policies that make use of these status values to restrict the registrant data shared with any or all clients according to the sensitivity of the data, the privileges of the clients, or some other heuristics. RDAP uses the jCard standard format for entity representation. Operators may find that many of the jCard fields are irrelevant for registry operation purposes or that they have no reason to collect information from registrants that would correspond to certain fields. Operators wishing to reduce privacy risks for registrants may restrict which information they collect and/or which fields they populate in responses. In addition to privacy risks to registrants, there are also potential privacy risks for those who query registration data. For example, the fact that a registry employee performs a particular query may reveal information about the employee's activities that he or she would have preferred to keep private. RDAP supports the use of HTTP over TLS to provide privacy protection for those querying registrant data as well as registrants, unless operational constraints make it impossible to meet this requirement.

One of the goals of RDAP is to provide security services that do not exist in the WHOIS protocol. This document describes the security services provided by RDAP and associated protocol layers, including authentication, authorization, availability, data confidentiality, and data integrity. Non-repudiation services were also considered and ultimately rejected due to a lack of requirements. There are, however, currently deployed WHOIS servers that can return signed responses that provide non-repudiation with proof of origin. RDAP might need to be extended to provide this service in the future. As an HTTP-based protocol, RDAP is susceptible to code injection attacks. Code injection refers to adding code into a computer system or program to alter the course of execution. There are many types of code injection, including SQL injection, dynamic variable or function injection, include-file injection, shell injection, and HTML-script injection, among others. Data confidentiality and integrity services provide a measure of defense against man-in-the-middle injection attacks, but vulnerabilities in both client- and server-side software make it possible for injection attacks to succeed. Consistently checking and validating server credentials can help detect man&nbhy;in&nbhy;the&nbhy;middle attacks. As noted in , digital certificates can be used to implement federated authentication. There is a risk of too promiscuous, or even rogue, CAs being included in the list of acceptable CAs that the TLS server sends the client as part of the TLS client-authentication handshake and lending the appearance of trust to certificates signed by those CAs. Periodic monitoring of the list of CAs that RDAP servers trust for client authentication can help reduce this risk. The Transport Layer Security protocol includes a null cipher suite that does not encrypt data and thus does not provide data confidentiality. This option MUST NOT be used when data confidentiality services are needed. Additional considerations for secure use of TLS are described in . Data integrity services are sometimes mistakenly associated with directory service operational policy requirements focused on data accuracy. "Accuracy" refers to the truthful association of data elements (such as names, addresses, and telephone numbers) in the context of a particular directory object (such as a domain name). Accuracy requirements are out of scope for this protocol. Additional security considerations are described in the specifications for HTTP , HTTP Basic and Digest access authentication , HTTP over TLS , and additional HTTP status codes . Security considerations for federated authentication systems can be found in the OAuth and OpenID specifications.

OpenID Foundation OASIS

The authors would like to acknowledge the following individuals for their contributions to this document: Richard Barnes, Marc Blanchet, Alissa Cooper, Ernie Dainow, Spencer Dawkins, Jean-Philippe Dionne, Byron Ellacott, Stephen Farrell, Tony Hansen, Peter Koch, Murray Kucherawy, Barry Leiba, Andrew Newton, and Linlin Zhou.