Sender Policy Framework

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Sender Policy Framework (SPF) is a simple email validation system designed to detect email spoofing by providing a mechanism to allow receiving mail exchangers to check that incoming mail from a domain is being sent from a host authorized by that domain's administrators.[1] The list of authorized sending hosts for a domain is published in the Domain Name System (DNS) records for that domain in the form of a specially formatted TXT record. Email spam and phishing often use forged sender addresses, so publishing and checking SPF records can be considered anti-spam techniques.

Sender Policy Framework is defined in IETF publication RFC 7208.

Principles of operation[edit]

Further information: Sender Rewriting Scheme (SRS)

The Simple Mail Transfer Protocol permits any computer to send email claiming to be from any source address. This is exploited by spammers who often use forged email addresses, making it more difficult to trace a message back to its sender, and easy for spammers to hide their identity in order to avoid responsibility. It is also used in phishing techniques, where users can be duped into disclosing private information in response to an email purportedly sent by an organization such as a bank.

SPF allows the owner of an Internet domain to specify which computers are authorized to send mail with sender addresses in that domain, using Domain Name System (DNS) records. Receivers verifying the SPF information in TXT records may reject messages from unauthorized sources before receiving the body of the message. Thus, the principles of operation are similar to those of DNS-based blackhole lists (DNSBL), except that SPF uses the authority delegation scheme of the Domain Name System. Current practice require the use of TXT records,[2] just as early implementations did. For a while a new record type (SPF, type 99) was registered and made available in common DNS software. Use of TXT records for SPF was intended as a transitional mechanism at the time. The experimental RFC, RFC 4408, section 3.1.1, suggested "An SPF-compliant domain name SHOULD have SPF records of both RR types".[3] The proposed standard, RFC 7208, says "Use of alternative DNS RR types was supported in SPF's experimental phase but has been discontinued".[2]

The sender address is transmitted at the beginning of the SMTP dialog. If the server rejects the sender, the unauthorized client should receive a rejection message, and if that client was a relaying message transfer agent (MTA), a bounce message to the original sending address may be generated. If the server accepts the sender, and subsequently also accepts the recipients and the body of the message, it should insert a Return-Path field in the message header in order to save the sender address. While the address in the Return-Path often matches other originator addresses in the mail header such as From or Sender, this is not necessarily the case, and SPF does not prevent forgery of these other addresses.

Spammers can send email with an SPF PASS result if they have an account in a domain with a sender policy, or abuse a compromised system in this domain. However, doing so makes the spammer easier to trace.

The main benefit of SPF is to the owners of e-mail addresses that are forged in the Return-Path. They receive large amounts of unsolicited error messages and other auto-replies. If such receivers use SPF to specify their legitimate source IP addresses and indicate FAIL result for all other addresses, receivers checking SPF can reject forgeries, thus reducing or eliminating the amount of backscatter.

SPF has potential advantages beyond helping identify unwanted mail. In particular, if a sender provides SPF information, then receivers can use SPF PASS results in combination with a white list to identify known reliable senders. Scenarios like compromised systems and shared sending mailers limit this use.

Reasons to implement[edit]

If a domain publishes an SPF record, spammers and phishers are less likely to forge e-mails pretending to be from that domain, because the forged e-mails are more likely to be caught in spam filters which check the SPF record. Therefore, an SPF-protected domain is less attractive to spammers and phishers. Because an SPF-protected domain is less attractive as a spoofed address, it is less likely to be blacklisted by spam filters and so ultimately the legitimate e-mail from the domain is more likely to get through.[4]

FAIL and forwarding[edit]

SPF breaks plain message forwarding. When a domain publishes an SPF FAIL policy, legitimate messages sent to receivers forwarding their mail to third parties may be rejected and/or bounced if all of the following occur:

  1. The forwarder does not rewrite the Return-Path, unlike mailing lists.
  2. The next hop does not white list the forwarder.
  3. This hop checks SPF.

This is a necessary and obvious feature of SPF – checks behind the "border" MTA (MX) of the receiver cannot work directly.

Publishers of SPF FAIL policies must accept the risk that their legitimate emails are being rejected or bounced. They should test (e.g., with a SOFTFAIL policy) until they are satisfied with the results. See below for a list of alternatives to plain message forwarding.

HELO tests[edit]

For an empty Return-Path as used in error messages and other auto-replies, an SPF check of the HELO-identity is mandatory.

With a bogus HELO identity the result NONE would not help, but for valid host names SPF also protects the HELO identity. This SPF feature was always supported as an option for receivers, and later SPF drafts including the final specification recommend to check the HELO always.

This allows receivers to white list sending mailers based on a HELO PASS, or to reject all mails after a HELO FAIL. It can also be used in reputation systems (any white or black list is a simple case of a reputation system).

Implementation[edit]

Compliance with SPF consists of three loosely related tasks:

Publish a policy 
Domains and hosts identify the machines authorized to send e-mail on their behalf. They do this by adding additional records to their existing DNS information: every domain name or host that has an A record or MX record should have an SPF record specifying the policy if it is used either in an email address or as HELO/EHLO argument. Hosts which do not send mail should have an SPF record published which indicate such ("v=spf1 -all"). It is highly recommended to validate the SPF record using record testing tools such as those provided on the SPF Project webpage.
Check and use SPF information 
Receivers use ordinary DNS queries, which are typically cached to enhance performance. Receivers then interpret the SPF information as specified and act upon the result.
Revise mail forwarding
Plain mail forwarding is not allowed by SPF. The alternatives are
  • remailing, i.e. replacing the original sender with one belonging to the local domain,
  • refusing, i.e. answering 551 User not local; please try <user@example.com>,
  • whitelisting on the target server, so that it will not refuse a forwarded message, and
  • Sender Rewriting Scheme, a more complicated mechanism that handles routing non-delivery notifications to the original sender.

Thus, the key issue in SPF is the specification for the new DNS information that domains set and receivers use. The records laid out below are in typical DNS syntax:

example.com. IN TXT "v=spf1 ip4:192.0.2.0/24 ip4:198.51.100.123 a -all"

"v=" defines the version of SPF used. The following words provide mechanisms to use to determine if a domain is eligible to send mail. The "ip4" and "a" specify the systems permitted to send messages for the given domain. The "-all" at the end specifies that, if the previous mechanisms did not match, the message should be rejected.

Mechanisms[edit]

Eight mechanisms are defined:

ALL Matches always; used for a default result like -all for all IPs not matched by prior mechanisms.
A If the domain name has an address record (A or AAAA) that can be resolved to the sender's address, it will match.
IP4 If the sender is in a given IPv4 address range, match.
IP6 If the sender is in a given IPv6 address range, match.
MX If the domain name has an MX record resolving to the sender's address, it will match (i.e. the mail comes from one of the domain's incoming mail servers).
PTR If the domain name (PTR record) for the client's address is in the given domain and that domain name resolves to the client's address (forward-confirmed reverse DNS), match. This mechanism is deprecated and should no longer be used.[2]
EXISTS If the given domain name resolves to any address, match (no matter the address it resolves to). This is rarely used. Along with the SPF macro language it offers more complex matches like DNSBL-queries.
INCLUDE If the included (a misnomer) policy passes the test this mechanism matches. This is typically used to include policies of more than one ISP.

Qualifiers[edit]

Each mechanism can be combined with one of four qualifiers:

  • + for a PASS result. This can be omitted; e.g., +mx is the same as mx.
  • ? for a NEUTRAL result interpreted like NONE (no policy).
  • ~ (tilde) for SOFTFAIL, a debugging aid between NEUTRAL and FAIL. Typically, messages that return a SOFTFAIL are accepted but tagged.
  • - (minus) for FAIL, the mail should be rejected (see below).

Modifiers[edit]

The modifiers allow for future extensions to the framework. To date only the two modifiers defined in the RFC 4408 have been widely deployed:

  • exp=some.example.com gives the name of a domain with a DNS TXT record (interpreted using SPF's macro language) to get an explanation for FAIL results—typically a URL which is added to the SMTP error code. This feature is rarely used.
  • redirect=some.example.com can be used instead of the ALL-mechanism to link to the policy record of another domain. This modifier is easier to understand than the somewhat similar INCLUDE-mechanism.

Error handling[edit]

As soon as SPF implementations detect syntax errors in a sender policy they must abort the evaluation with result PERMERROR. Skipping erroneous mechanisms cannot work as expected, therefore include:bad.example and redirect=bad.example also cause a PERMERROR.

Another safeguard is the maximum of ten mechanisms querying DNS, i.e. any mechanism except from IP4, IP6, and ALL. Implementations can abort the evaluation with result SOFTERROR when it takes too long or a DNS query times out, but they must return PERMERROR if the policy directly or indirectly needs more than ten queries for mechanisms. Any redirect= also counts towards this processing limit.

A typical SPF HELO policy v=spf1 a -all may execute up to three DNS queries: (1) TXT, (2) SPF (obsoleted by RFC 7208), and (3) A or AAAA. This last query counts as the first mechanism towards the limit (10). In this example it is also the last, because ALL needs no DNS lookup.

Caveats[edit]

Interpretation[edit]

SPF FAIL policies can be an effective but problematic tool. A typical example is a user that wishes to send an email from a private PC or a mobile phone: the user uses their corporate email address but may use a different outgoing SMTP server which is not listed in the SPF record. The corporate domain may therefore be secure by blocking all email that does not originate from themselves, but have thereby limited some of their own users. Many organizations consider this compromise acceptable and even desirable.

SPF PASS is useful for authenticating the domain for use as a parameter to a spam classification engine. That is, the domain in the sender address can be considered to be authentic if the originating IP yields an SPF PASS. The domain can then be referenced against a reputation database.

SPF results other than PASS (used in combination with a reputation system) and FAIL cannot be meaningfully mapped to PASS and FAIL. However, a reputation system can easily track independent reputations for each SPF result, i.e. example.com:PASS and example.com:NEUTRAL would have different reputations, and ditto for the other results. This approach is useful even without whitelisting plain forwarders, because the FAIL results from the plain forwarders simply accrue an independent reputation.

The meaning of PASS, SOFTFAIL, FAIL is sometimes incorrectly interpreted to mean "not-spam", "maybe-spam", "spam" respectively. However SPF does nothing of the sort. SPF merely offers an organization firstly the means to classify emails based on their domain name instead of their IP address (SPF PASS); and secondly, the means to block unauthorized use of their domain (SPF FAIL).

Intra-domain forgery[edit]

In a naive implementation, SPF does not prevent a user with the same domain sending an email on behalf of another user because only the domain part of the address is used to locate the SPF policy record. In more sophisticated implementations, the domain owner can specify separate policies for each user by means of SPF "macros" that reference the "localpart" (user) as defined in RFC 4408, or simply require all mail submissions for the domain to use SMTP AUTH (RFC 4954). The latter is highly recommended anyway for many reasons.

Checkpoints[edit]

SPF needs to operate on the host indicated by the receiving domain's MX record. This means the host(s) that are the direct recipient of remote TCP connections, because such a host can easily deduce the originating IP address from the TCP session. These hosts are able to block the email during the SMTP session, avoiding the necessity to generate bounce messages which could be backscatter. (As the SPF RFC 4408 says, "Generating non-delivery notifications to forged identities that have failed the authorization check is generally abusive and against the explicit wishes of the identity owner.")

Other downstream hosts, for instance in a forwarding scenario, can only perform SPF checks based on "Received" headers. This is cumbersome and error-prone. A better approach is for the MX host to check SPF without blocking any email, and then add a "Received-SPF" header field as specified in RFC 4408 or the newer "Authentication-Results" header of RFC 7001. Downstream hosts can then look at these trace headers and set their own policy of whether to reject, accept, or quarantine based on the SPF result and other factors.

DoS attack[edit]

An Internet draft[5] discussed concerns related to the scale of an SPF answer leading to network exploits as a means to corrupt the DNS. This issue is also covered in the security considerations of the SPF RFC. The SPF project did a detailed analysis[6] of this draft and claimed that SPF does not pose any unique threat of DNS DoS, citing example attacks using NS and MX records and identifying void DNS lookups (negative caching) as the key DNS weakness.

An SPF based attack can generate more than 40KB of traffic per message originating completely from recipient resources once a spam campaign containing MAILFROM with unique local-parts exceed the recipient's negative caching limits commonly imposed. Often Windows based services impose rather low limits and other resolver offerings also permit low negative caching limits to mitigate access problems following service disruptions. SPF includes the "l" macro able to combine components of the email-address local-parts to construct unique DNS requests generated by recipient resources. The SPF result for "jo@example.com" may not be the same as that for "lu@example.com" which always requires repeating potentially long sequences of more than 100 DNS transactions based upon the same cached SPF record.

SPF provides several advantages for malefactors:

  1. SPF based attacks reflect off recipient resources which obfuscates likely compromised systems initiating the abuse where Authentication-Results omit the authorized IP addresses.
  2. SPF based attacks, even without short negative caching, offers a means to obtain significant network amplification.
  3. SPF based attacks are seldom logged.
  4. When based upon the MAILFROM, the common case, authentication is not achieved so referenced domain's involvement may not extend beyond mere authorization.

Relationship with DKIM[edit]

SPF validates the message envelope (the SMTP bounce address), not the message contents (header and body) – this is the distinction between SMTP (as specified in STD 10 or RFC 5321) and Internet Message Format (as specified in STD 11 or RFC 5322). It is orthogonal and complementary to DomainKeys Identified Mail (DKIM), which signs the contents (including headers).

In brief, SPF validates MAIL FROM vs. its source server; DKIM validates the "From:" message header and a mail body by cryptographic means.

Sender ID[edit]

Sender ID RFC 4406, is a parallel solution to the problem of message validation, and defines a pair of closely related tests. One validates a message's Purported Responsible Address (PRA) as defined in RFC 4407. The other validates a message's Reverse-Path (also known as MAIL-FROM address) as defined in RFC 4408.

Quoting from RFC4407:

"Note that the Sender ID experiment may use DNS records that may have been created for the current SPF experiment or earlier versions in this set of experiments. Depending on the content of the record, this may mean that sender-ID heuristics would be applied incorrectly to a message. Depending on the actions associated by the recipient with those heuristics, the message may not be delivered or may be discarded on receipt."

Those publishing SPF DNS records should consider the advice given in section 3.4 of RFC 4406 and may wish to publish both v=spf1 and spf2.0 records to avoid the conflict. However, note that despite the version label used, Sender ID is not technically a second revision of SPF.

Caveats:

  • Because PRA is defined as a fall-back hierarchy, the only header that is actually protected is "Resent-Sender". Any other header can be forged by simply adding a higher-priority header with a different domain (the highest priority being Resent-Sender). DKIM cryptographically protects an arbitrary number of header fields.
  • Mail with a Return-Path failing in an SPF check can be rejected during the SMTP envelope phase, before bandwidth has been wasted on the actual email. Because SMTP does not provide a way to reject an email after the headers, but before the body, the entire email must be received before PRA can be checked.

Wide-mask vulnerability[edit]

Some spammers use SPF to decrease spam-rating by specifying wide mask in valid server address, so any spam from botnets becomes spf-valid and the probability to pass spam-filters increases.

seminar-for-you.ru.        14400   IN      TXT     "v=spf1 a mx ip4:55.11.65.20/2 ip4:90.2.123.112/2 ip4:176.33.87.19/2 ip4:212.63.89.33/2 -all"
worldwidemail.ru.  13733   IN      TXT     "v=spf1 a mx ip4:55.11.65.20/2 ip4:90.2.123.112/2 ip4:176.33.87.19/2 ip4:212.63.89.33/2 -all"
example.com.               21600   IN      SPF     "v=spf1 +all"

This last record says that any host on the Internet may send mail on behalf of the domain/hostname example.com. Although syntactically valid, "+all" is indicative of an administrator who does not care about SPF or the mail forgeries it detects.

For stable domains, this simply means that any reputation attached to the domain is the same with or without SPF and such spam domains are easily learned and rejected. The real value of wide-mask SPF policies to spammers is with "throw-away" domains that are registered, used to send spam from botnets for a few days, and then abandoned.

History[edit]

The first public mention of the concept was in 2000 but went mostly unnoticed.[7] No mention was made of the concept again until a first attempt at an SPF-like specification was published in 2002 on the IETF "namedroppers" mailing list by Dana Valerie Lank (previously D. Green), who was unaware of the 2000 mention of the idea. The very next day, Paul Vixie posted his own SPF-like specification on the same list. These posts ignited a lot of interest, and eventually led to the forming of the IETF Anti-Spam Research Group (ASRG) and their mailing list, where the SPF idea was debated among a subscriber base that seemed to grow exponentially day by day. Among the proposals submitted to the ASRG were "Reverse MX" by Hadmut Danisch, and "Designated Mailer Protocol" by Gordon Fecyk.[8]

In June 2003, Meng Weng Wong merged the RMX and DMP specifications[9] and solicited suggestions from other programmers. Over the next six months, a large number of changes were made and a large community had started working on SPF.[10]

Originally SPF stood for Sender Permitted From and was sometimes also called SMTP+SPF, but its name was changed to Sender Policy Framework in February 2004.

In early 2004, the IETF created the MARID working group and tried to use SPF and Microsoft's CallerID proposal as the basis for what is now known as Sender ID.

After the collapse of MARID the SPF community returned to the original "classic" version of SPF. In July 2005 this version of the specification was approved by the IESG as an IETF experiment, inviting the community to observe SPF during the two years following publication. On April 28, 2006, the SPF RFC was published as experimental RFC 4408.

Controversy[edit]

In 2004, Steven M. Bellovin wrote an e-mail that discusses his concerns with SPF.[11] Some of these include:

  • SPF originally used TXT records in DNS, which are supposed to be free-form text with no semantics attached. SPF proponents readily acknowledge that it would be better to have records specifically designated for SPF, but this choice was made to enable rapid implementation of SPF. In July 2005, IANA assigned the Resource Record type 99 to SPF. Later on, the use of SPF records was discontinued, and it is required to use TXT records to this day.[2]
  • As of the time he wrote his message, there was no consensus that this is the right way to go. Some major e-mail service providers have not bought into this scheme. Unless and until they do, it does not help much, either for their customers (who make up a substantial proportion of the user population) or for everyone else (because their addresses could be forged). It is worth noting that since this concern was raised, Google Mail and AOL, among others, have embraced SPF.[12]
  • Bellovin's strongest concerns involve the underlying assumptions of SPF (SPF's "semantic model"). When using SPF, the SPF DNS records determine how a sender is allowed to send. That means that the owner of the domain will control how senders are allowed to send. People who use "portable" e-mail addresses (such as e-mail addresses created by professional organizations) will be required to use the domain owner's SMTP sender, which may not currently even exist. Organizations providing these "portable" addresses could, however, create their own mail submission agents (MSAs) (RFC 6409) or offer VPNs or simply not publish an SPF record. Besides, SPF only ties the SMTP Return-Path to permitted MSAs; users are still free to use their RFC 5322 addresses elsewhere.

There are other concerns about the impact of widespread use of SPF, notably the impact on various legitimate forms of email spoofing,[13] such as forwarding services, SMTP use by people with multiple identities, etc. (For example, a person who uses their home ISP's SMTP servers to send mail with their work email as the address.) On the other hand, many of these uses may be "expected" yet not "legitimate". To a certain extent this is more a question of ownership and expectations than a technical question.

The IETF spfbis working group, tasked with reworking the SPF specification aiming for "Proposed Standard" status in a new RFC, during April 2013 appeared to have reached consensus around deprecating SPF type 99 in favour of continued TXT record usage.[14] People from the DNSEXT working group opposed this strongly in a series of email threads on spfbis, dnsext, and IETF general discussion mailing lists.[15][16] The spfbis working group chair requested to stop that torrent of protest, since the discussion on the resource record type (RRTYPE) in the working group was terminated long time before [17] a move that was seen as trying to silence the protests by some fierce DNS purists. An independent draft was proposed later, documenting how the spurious recursion to TXT records is characterized in the current Internet.[18]

Deployment[edit]

Anti-spam software such as SpamAssassin version 3.0.0 and ASSP implement SPF. Many mail transfer agents (MTAs) support SPF directly such as Courier, CommuniGate Pro, Wildcat, MDaemon, and Microsoft Exchange, or have patches/plug-ins available that support SPF, including Postfix, Sendmail, Exim, qmail, and Qpsmtpd.[19] As of 2013, more than seven million domains publish SPF FAIL -all policies.[20]

In August 2005 it was learned that EarthLink would refuse to allow hosted domains the ability to enter SPF records.[21]

In a survey published in 2007, 5% of the .com and .net domains had some kind of SPF policy. In 2009, a continuous survey run at Nokia Research reports that 51% of the tested domains specify an SPF policy.[22] These results can include trivial policies like v=spf1 ?all.[23] In April 2007, BITS, a division of the Financial Services Roundtable, published e-mail security recommendations for its members including SPF deployment.[24]

In 2008, the Messaging Anti-Abuse Working Group (MAAWG) published a paper about email-authentication covering SPF, Sender ID, and DKIM.[25] In their "Sender Best Communication Practices" the MAAWG stated: "At the very least, senders should incorporate SPF records for their mailing domains".[26]

See also[edit]

References[edit]

  1. ^ "Sender Policy Framework: Introduction". 
  2. ^ a b c d Scott Kitterman (April 2014). "DNS Resource Records". Sender Policy Framework (SPF) for Authorizing Use of Domains in Email, Version 1. IETF. sec. 3.1. RFC 7208. https://tools.ietf.org/html/rfc7208#section-3.1. Retrieved 26 April 2014.
  3. ^ Wong, M., and W. Schlitt. RFC 4408. April 2006 <http://tools.ietf.org/html/rfc4408>
  4. ^ "Why should I implement a SPF record on my domain?". Email Manual. May 2009. Retrieved 2010-01-01. [dead link]
  5. ^ SPF DoS Exploitation (expired Internet draft)
  6. ^ OpenSPF analysis of the DoS attack draft
  7. ^ "SPF: First Public Mention 2000". Retrieved 28 August 2014. 
  8. ^ "SPF: History/Pre-SPF". Retrieved 16 May 2009. 
  9. ^ For a comparison among RMX, DMP and SPF, see RMX and DMP compared on the historical openspf site.
  10. ^ "SPF: History/SPF-2003". Retrieved 16 May 2009. 
  11. ^ Steve Bellovin expresses doubts (Jan 2004)
  12. ^ "SPF Information". AOL Postmaster. Archived from the original on 2007-07-08. Retrieved 2007-10-04. 
  13. ^ "Problems with Designated Sender". Taughannock Networks. Retrieved 2009-12-16. 
  14. ^ Murray Kucherawy (July 212). "Conclusions". Resolution of the Sender Policy Framework (SPF) and Sender ID Experiments. IETF. sec. 6. RFC 6686. https://tools.ietf.org/html/rfc6686#section-6. Retrieved 16 December 2013.
  15. ^ S Moonesamy (Tue, 23 Apr 2013 15:03:45 -0700). "Obsoleting SPF RRTYPE". DNSEXT Discussion List. IETF. Retrieved 16 December 2013.  Check date values in: |date= (help)
  16. ^ Dan Schlitt (Thu, 29 Aug 2013 12:30:07 -0400). "Last Call: <draft-ietf-spfbis-4408bis-19.txt> (Sender Policy Framework (SPF) for Authorizing Use of Domains in Email, Version 1) to Proposed Standard". IETF Discussion List. IETF. Retrieved 16 December 2013.  Check date values in: |date= (help)
  17. ^ Andrew Sullivan (Wed, 29 May 2013 17:42:34 -0400). "The RRTYPE topic". SPFBIS Discussion List. IETF. Retrieved 16 December 2013.  Check date values in: |date= (help)
  18. ^ John Klensin; Andrew SUllivan; Patrik Fältström (August 2013). An IANA Registry for Protocol Uses of Data with the DNS TXT RRTYPE. IETF. I-D draft-klensin-iana-txt-rr-registry. https://tools.ietf.org/html/draft-klensin-iana-txt-rr-registry. Retrieved 16 December 2013.
  19. ^ "Qpsmtpd SPF plugin". 2013. 
  20. ^ "SPF -all Domain Survey". 2013. Retrieved 2013-04-23. 
  21. ^ "SPF Loses Mindshare". 2005. Retrieved 4 April 2011. 
  22. ^ "Nokia Research Report on SPF Adoption". 
  23. ^ Liu, Cricket (January 2007). "Handicapping New DNS Extensions and Applications". ONLamp. Retrieved 2007-10-04. 
  24. ^ "BITS Email Security Toolkit" (PDF). BITS. April 2007. Retrieved 2008-06-13. 
  25. ^ Crocker, Dave (March 2008). "Trust in Email Begins with Authentication" (PDF). MAAWG. Retrieved 2011-07-28. 
  26. ^ "MAAWG Sender Best Communications Practices Executive Summary" (PDF). MAAWG. 2011-10-07. Retrieved 2012-04-27. 

External links[edit]