|Network Working Group||D. Crocker|
|INTERNET DRAFT||Brandenburg InternetWorking|
|Category: Standards Track||JLC.net|
|Expires: August 2005||D. Otis|
|Mail Abuse Prevention System|
Certified Server Validation (CSV)
This document is an Internet-Draft and is subject to all provisions of section 3 of RFC 3667. By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she become aware will be disclosed, in accordance with RFC 3668.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress".
The list of current Internet-Drafts can be accessed at <http://www.ietf.org/ietf/1id-abstracts.txt>.
The list of Internet-Draft Shadow Directories can be accessed at <http://www.ietf.org/shadow.html>.
This Internet-Draft will expire in August 2005.
Copyright (C) The Internet Society (2005). All Rights Reserved.
Internet mail relies on exchanges between systems that have made no prior arrangement with each other. Widespread abuse of the email system has led operators to demand accountability for the email their receiving SMTP servers are being asked to process. Certified Server Validation (CSV) provides an economical service that permits a receiving SMTP server to decide whether a sending SMTP client is likely to produce well-behaved traffic, or at least to decide whether the client is sufficiently accountable for its actions. CSV provides a small, simple and useful improvement to Internet mail service accountability. It builds upon the existing practise of service providers that accredit the networks from which sending systems are connecting.
CSV considers two questions at the start of each SMTP session:
To validate an SMTP session from an unknown sending SMTP client using CSV, a typical sequence for the receiving SMTP server is:
Using whatever thresholds are set by the receiving site's policies:
Internet mail suffers from the operation of hosts acting as mail transfer agents (MTA) without any meaningful cross-net accountability. This makes it impossible to vet MTAs or find recourse when their operations cause problems. Many of these hosts have been compromised and have been turned into unwilling participants in large networks of hostile MTAs that send spam and worms, and contribute to denial of service attacks.
When a server MTA receives a connection, it decides whether to accept the message traffic that is being sent to it, trusting that its delivery will not be problematic to the operation of the provider or their users. How can it do this, when operating in the open Internet? Certified Server Validation (CSV) defines a service that permits the receiving SMTP server to decide whether messages sent by the sending SMTP client are likely to be well-behaved, or at least to decide whether that client is sufficiently accountable for its actions.
The process of deciding on this trust of the client requires performing a series of conceptually discrete steps:
A proposal or its implementation well might combine some of these steps. However it is important to consider them independently, in order to ensure that the proposal specifies that they are performed in a valid manner, or at least that the constraints of the proposal are clear for each of these conceptual functions. This specification distinguishes each of these logical steps and defines their operation separately. It is based on validation of the EHLO domain name. The proposed mechanism is small, simple and useful. In particular it permits detecting machines that are prohibited from acting as Client MTAs and those that are permitted. The mechanism is designed to be useful between peer MTAs and only requires use of well-established mechanisms.
CSV builds upon this popular model. Besides the considerable benefit of having operational practice, the model can be extremely efficient. It permits the service provider to assess the source of an entire message stream, rather than having to evaluate each message. Also, CSV makes its assessment before messages cross the Internet, thereby saving bandwidth and reducing the impact of a distributed denial of service attack.
CSV verifies that a host is authorized to act as an SMTP client and that the client is likely to be operated acceptably. CSV enhances current practice with:
For a receiving SMTP server to determine whether a host has authorization to act as a sending SMTP client, it is necessary to identify the host and verify its association with that identity. Given that, a DNS query on the name can return an explicit authorization.
The means of identifying a remote host or service requires uniqueness and is aided by persistence. The identifier must not be ambiguous and its use is made far more efficient if it is stable over time. The two usual choices are IP Addresses and Domain Names.
An IP Address typically refers to a single host and can change relatively frequently, as the host's connection to the Internet changes. IP Addresses are reported by the Internet infrastructure and for simple security requirements, transactional use of an IP Address through the Internet's routing fabric is taken as validation of the Address.
Domain Names are longer-lived but require new administrative effort. They can be used to refer to multiple hosts simultaneously. The DNS administrator for a domain will maintain record(s) listing one or more IP addresses associated with that name, even though reverse-DNS records (not controlled by the same DNS administrator) may give conflicting information. The forward-DNS is considered the valid authority for CSV purposes. Therefore, authentication of a domain name's reference to a particular IP Address requires an explicit authentication step.
If the sending SMTP client of a connection can be authenticated, then it is possible to develop an accountability mechanism based on that authentication. MUA-MSA exchanges have a substantial number of useful authentication mechanisms available. These are often very strong, and involve significant prior arrangement. The same holds true for MDA-MUA exchanges, and often for MSA-MTA and MTA-MDA exchanges, such as within an organization's local network.
What is missing is a useful means of authenticating MTA-MTA exchanges over the open Internet. Prior arrangement between such a pair of MTAs is antithetical to the history and operation of Internet mail. Spontaneous communications are at the core of Internet design and operation. So the challenge is to develop an authentication mechanism that permits the necessary amount of accountability, without imposing undue overhead or restrictions.
A number of strong authentication mechanisms are possible, but none has yet attained widespread adoption among MTAs with no prior relationship. CSV specifies a weaker authentication scheme that meets the modest requirements for this service. Stronger methods can be supported later, if necessary. However they must be tied to a domain name and must not require any prior relationship.
Internet operation has typically required no public mechanism for restricting or permitting particular hosts to operate clients or servers for particular services on behalf of particular domains. The DNS MX record states where to route email that is destined for a specific domain; this implies a degree of authorization for the host referenced in the MX. However the record is really for routing and there is no equivalent means of specifying authorization of other hosts that might act as email relays. Similarly there is no means for checking the authorization of World Wide Web servers, DNS servers, telnet clients or other Internet applications.
What is missing is an open, interoperable means by which accountable domain management can announce its authorization of a particular host to operate a particular service. CSV defines such a mechanism for sending SMTP clients.
This portion of CSV determines accreditations for the sending SMTP client or for the administration under which it operates. The basis for deciding that an authorizing agency is, itself, to be trusted can be highly varied. Often, well-established practices are not that well-understood. This makes it difficult to predict what methods of accreditation will be most appropriate and successful for Internet mail. It is expected that this portion of an Internet mail validation service will therefore need to support be a variety of accreditation service styles.
What is needed is a standard means for:
CSV defines a mechanism for session-time, domain-based validation of a sending SMTP client. It is useful across the open Internet, between MTAs that have made no prior arrangement with each other. Validation establishes that the operation of the MTA is authorized by an accredited administrator of the declared domain name.
The validation requirements are modest, because the system does not seek to provide long-term vetting of the client host, nor does it assess the actual content being exchanged. Techniques that would be wholly inadequate for classic, strong authentication and validation can be entirely sufficient for CSV's needs.
Validation has two separate phases: assessing authorization and assessing accreditation. The first is performed between the receiving SMTP server and the sending SMTP client. The second is performed between the receiving SMTP server and one or more accrediting services.
This phase provides a means for a network administrator to publicly state what hosts are authorized by it to act as client MTAs. Absent such a statement of authority, it is possible that the client is a rogue or compromised host.
The assessment requires three steps:
The CSV authorization phase provides a basis for trusting that the sending SMTP client is under the control of a domain's management; but this says nothing about the policies and practices of that management. Separate accreditation services are needed for that. It is expected that there will be numerous services that provide accreditation. CSV is intended to support use of any service that gains credibility among operators of SMTP servers.
One form of accreditation service is particularly easy to use initially: a private list, maintained by the user of the information. That is, a receiving SMTP server can manage its own, private list of trusted domains. This is not viable for the long-term, given the number of possible, valid client MTAs and the rate of on-going changes.
Long term use is expected employ queries to independent, third-party services. CSV provides a set of capabilities for using external accreditation. (See "Domain Name Accreditation" in [ID-CSVDNA].) Sending SMTP clients SHOULD publish CSV records referring to accreditation services in which they are listed. Accreditation services MUST publish DNA-conformant records.
CSV defines a security mechanism. The nature of the security requirements for CSV are significantly different from typical, "strong" methods required for most Internet security functions.
The proposal relies on the integrity and authenticity of DNS data.
|[ID-CSVCSA]||Otis, D., Crocker, D. and J. Leslie, "sending SMTP client Authorization (CSA)", June 2004.|
|[ID-CSVDNA]||Leslie, J., Crocker, D. and D. Otis, "Domain Name Accreditation (DNA)", June 2004.|
|[RFC0791]||Postel, J., "Internet Protocol", STD 5, RFC 791, September 1981.|
|[RFC0821]||Postel, J.B., "Simple Mail Transfer Protocol", STD 10, RFC 821, August 1982.|
|[RFC0822]||Crocker, D.H., "Standard for the format of ARPA Internet text messages", STD 11, RFC 822, August 1982.|
|[RFC1035]||Mockapetris, P., "Domain names - implementation and specification", STD 13, RFC 1035, November 1987.|
|[RFC1122]||Braden, R., "Requirements for Internet Hosts - Communication Layers", STD 3, RFC 1122, October 1989.|
|[RFC2554]||Myers, J.G., "SMTP Service Extension for Authentication", RFC 2554, March 1999.|
|[RFC2782]||Gulbrandsen, A., Vixie, P. and L. Esibov, "A DNS RR for specifying the location of services (DNS SRV)", RFC 2782, February 2000.|
|[RFC2821]||Klensin, J., "Simple Mail Transfer Protocol", RFC 2821, April 2001.|
|[RFC2822]||Resnick, P., "Internet Message Format", RFC 2822, April 2001.|
|[RFC3207]||Hoffman, P., "SMTP Service Extension for Secure SMTP over Transport Layer Security", RFC 3207, February 2002.|
|[ID-brand-drip]||Brand, R and L Sherzer, "Designated Relays Inquiry Protocol (DRIP)", Internet-Draft draft-brand-drip-02, October 2003.|
|[ID-mail-arch]||Crocker, D., "Internet Mail Architecture", May 2004.|
|675 Spruce Drive
|Sunnyvale, CA 94086|
|10 Souhegan Street
|Milford, NH 03055|
|Mail Abuse Prevention System|
|1737 North First Street, Suite 680
|San Jose, CA 94043|
This proposal is similar to DRIP [ID-brand-drip], however it uses a different DNS [RFC1035] record.
Review comments and suggestions, on previous versions of CSV, have been made by: Tony Finch, Carl Hutzler, Meng Weng Wong, Greg Connor.
The routing infrastructure of the Internet distinguishes hosts by their topological attachment, noted as its IP Address. Because IP Addresses change periodically and users prefer references that can be mnemonic, hosts on the Internet generally have one or more Domain Names (DNS) [RFC1035] assigned to them. A Domain Name is globally unique. The core function of the DNS is to map from a name supplied by the user, to an IP Address associated with that name. Internet protocols often permit a host to identify itself with its domain name.
But what if a host is programmed incorrectly, or even maliciously. We need a way to authenticate that a host is reporting its name correctly. Establishing this authentication is separate from determining its authorization to perform any particular service. Until the relationship is authenticated, we cannot apply policies associated with the name.
A number of methods for authenticating the relationship between the host and its reported name might be used. The current CSV specification supports authentication through Domain Name Service mappings between a domain name and an IP Address. Other equally valid methods are possible. However none has yet proved practical for authenticating a client to a server, without prior arrangement between them.
The Domain Name System has a common mapping mechanism that can be used in a variety of ways, based on the schema for assigning names and the types of data listed under those names. The two most popular schemas are forward mapping and Reverse-DNS. Forward looks up a "regular" domain name and receives information about it, such as a list of IP Addresses associated with that name. Reverse DNS starts with an IP Address and maps it to a pointer to a "regular" domain name.
Often when contacted by a remote host, a host uses a reverse-DNS query to get the name of the remote host. This can be followed by a forward-DNS query to see if the name reported by the reverse-DNS query matches an IP address reported by the forward-DNS query. If so, this is generally considered an authentication of the relationship of the name to the host. This method is often used by receiving SMTP servers to decide whether to trust the sending SMTP client.
Closing the circle in this manner permits verifying both that the domain assigning the name and the service provider assigning IP addresses agree that this is the appropriate name for that remote host. Although this process has known limitations, it is considered sufficient for many basic uses.
Use of an IP Address returned by the DNS is sufficient for CSV-related authentication requirements of this service. However it MUST NOT be considered a strong form of authentication as to allow otherwise privileged access. The use of this mechanism is to aid selection of accreditation services, such as whether to query using the domain name or the client address. Other measures may be taken intended to limit exposure to unknown clients but are beyond the scope of this specification.
Reverse DNS can be used by itself to associate a domain name with an IP address. It indicates that the entity responsible for allocating that block of IP addresses has designated an IP address to be used by the domain name. Unfortunately, the reverse-IP branch of the DNS has a long history of being poorly maintained, and often does not match the forward-DNS information even when the relationship of host to name is genuine.
Reverse DNS by itself SHOULD NOT be considered sufficient authentication.
An isolated forward lookup is sufficient for simple sending SMTP client authentication, if an IP Address returned for that name matches the IP Address reported by the underlying IP service for that remote host. This indicates that the domain in question currently designates that IP Address as an IP address entitled to respond for that domain name.
The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79.
Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at <http://www.ietf.org/ipr>.
The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at email@example.com.
This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
Copyright (C) The Internet Society (2005). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights.
Funding for the RFC Editor function is currently provided by the Internet Society.