| TOC |
|
The objective of this document is to describe the design of so-called DNS-based malware protection services deployed today by Internet Service Providers (ISPs), DNS Application Service Providers (ASPs), and other organizations providing so-called DNS-based malware protection services via their recursive DNS services, as well as to describe the recommended practices regarding such systems. This document specifically and narrowly addresses those cases where DNS Redirect is being utilized to provide a service for end users which blocks domains hosting malicious software.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/.
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.”
This Internet-Draft will expire on March 8, 2011.
Copyright (c) 2010 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.
1.
Requirements Language
2.
Introduction
3.
Document Scope
4.
Terminology
5.
Malicious Site Protection
6.
Opt-In or Opt-Out Mechanisms
7.
Practices to Avoid
8.
Functional Design
8.1.
DNS Recursive Resolver
8.2.
Web Error Landing Server
8.3.
Web Browser Client
8.4.
Domain White List
8.5.
Malicious Domain List
8.6.
Legally-Mandated DNS Redirect Domain List
8.7.
Content-Based DNS Redirect Domain List
9.
Example DNS and HTTP Flows
10.
DNSSEC Considerations
11.
Security Considerations
12.
IANA Considerations
13.
Contributors
14.
Acknowledgements
15.
Normative References
Appendix A.
Document Change Log
Appendix B.
Open Issues
§
Authors' Addresses
| TOC |
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 [RFC2119] (Bradner, S., “Key words for use in RFCs to Indicate Requirement Levels,” March 1997.).
| TOC |
Internet users typically are provided with several IP addresses for recursive DNS servers, as described in Section 2.3 of [RFC1591] (Postel, J., “Domain Name System Structure and Delegation,” March 1994.), by their respective ISPs, typically in an automated fashion via DHCP [RFC2131] (Droms, R., “Dynamic Host Configuration Protocol,” March 1997.). Some other users and organizations choose to use a different set of IP address for their DNS servers, which are hosted and managed by another organization, such as a DNS ASP. It is also the case that a number of users and organizations choose to operate their own DNS servers, though those use cases are outside of the scope of this document.
ISPs and DNS ASPs have discovered over time that their users would like " enhanced " DNS services which can protect those users from reaching domains or fully qualified domain names (FQDNs, Section 5.1 of [RFC1035] (Mockapetris, P., “Domain names - implementation and specification,” November 1987.)) that would cause a user to inadvertently access malicious software, otherwise known simply as malware.
This document describes the design and function of a DNS-based malware protection service which only provides protection from domains hosting malware, as well as recommended practices and practices to avoid.
| TOC |
This document focuses on the systems and practices of ISPs and DNS ASPs. All other use cases, such as when an Internet user or organization chooses to operate their own DNS servers is outside of the scope of this document.
There are several ways that such entities can provide users with these enhanced DNS services. In addition to methods which rely primarily upon a recursive DNS server, alternate methods include (a) interception and replacement of a malware-hosting domain or FQDN by web browser client software, (b) interception and replacement of a malware-hosting domain or FQDN by a tool bar, plug-in, personal firewall security software or other web browser client add-on. These alternate methods, which rely upon various types of client software, are also outside of the scope of this document.
It is important to note that while these alternate methods are considered out of scope for this document, this should not be interpreted as a negative judgment of their suitability or applicability to the relevant problem space. Instead, these should simply be considered as alternate methods since, as with most any technical problem, there are a variety of valid methods for solving a problem.
While the Section 6 (Opt-In or Opt-Out Mechanisms) section indicates that users must be able to opt into or out of DNS-based malware protection services, the reasons for why an ISP or DNS ASP may choose one or the other as the default are out of scope.
Lastly, in the Section 5 (Malicious Site Protection) section of this document, the method by which FQDNs, domains, and/or sites are added or removed from malware lists is outside the scope of this document. [EDITORIAL NOTE: THIS MAY CHANGE IN A FUTURE VERSION OF THE DOCUMENT]
| TOC |
While these terms are generally well known, it is important to define them in the context of this document.
| TOC |
An Internet Service Provider, which provides Internet services, including basic network connectivity. It is not germane to this document what the method of connection is, such as wired or wireless, what the speed of such a connection is, or what other services are included or available to users. It is, however, assumed that the ISP is providing recursive DNS services to their users and is in some manner providing users with the IP addresses of these DNS servers, whether via DHCP, static assignment by users, or some other method.
| TOC |
A DNS Application Service Provider, which provides managed and/or hosted recursive DNS services (and possibly other DNS services) to their users. In the case of managed services, the DNS ASP may remotely manage the recursive DNS servers in a user's network. For a hosted recursive DNS service, these servers are typically located outside of the user's network and these hosted resources are shared across multiple users. In most instances, these are hosted services and users are manually configuring either their DHCP server or their individual computing devices with the IP addresses of the recursive DNS servers operated by their ASP.
| TOC |
An Internet user, which is generally a person using a computing device to connect to and make use of the Internet. Such users are typically connected at the edge of the network, though the method by which they connect to the Internet is not particularly relevant to this document.
| TOC |
A DNS recursive resolver processes fully qualified domain name queries (FQDN, Section 5.1 of [RFC1035] (Mockapetris, P., “Domain names - implementation and specification,” November 1987.)) into IP addresses by finding the resource records in the authoritative DNS servers for the domain associated with the FQDN. The resource records are then cached on the recursive server for future requests until an expiration timer expires called time to live (TTL), as described in Section 5.2 of [RFC2181] (Elz, R. and R. Bush, “Clarifications to the DNS Specification,” July 1997.). These servers are in most cases provided by ISPs for name resolution.
| TOC |
Client software operated by the user locally on their computing device, such as Microsoft Internet Explorer, Mozilla Firefox, Apple Safari, Google Chrome, etc.
| TOC |
The web server that a user's web browser is directed to when the DNS Recursive Server matches a DNS query to a malicious domain or FQDN. The contents of the web page that the web server sends the user varies widely across different ISPs and DNS ASPs. In most cases it simply explains that the attempted URL contains malware and that access has been prevented, though there are many other possibilities.
| TOC |
The web server that a user is directed to via a link on a page served by the Web Error Landing Server, the Malicious Domain Web Error Landing Server, from another system such as an account management system, or via direct access, which enables a user to control whether or not they are opted into or opted out of DNS-based malware protection services. This is described in additional detail in the Section 6 (Opt-In or Opt-Out Mechanisms) section.
| TOC |
In this document, an NXDOMAIN (nonexistent domain) response can be used interchangeably with an RCODE 3 response. The RCODE 3 response was first documented in see Section 4.1.1 of [RFC1035] (Mockapetris, P., “Domain names - implementation and specification,” November 1987.)). Subsequent RFCs introduced the term NXDOMAIN response, which is synonymous with RCODE 3 and tends to be used more frequently, as noted in Section 2.2 of [RFC2136] (Vixie, P., Thomson, S., Rekhter, Y., and J. Bound, “Dynamic Updates in the Domain Name System (DNS UPDATE),” April 1997.), Section 1 of [RFC2308] (Andrews, M., “Negative Caching of DNS Queries (DNS NCACHE),” March 1998.), and Section 5.4 of [RFC2535] (Eastlake, D., “Domain Name System Security Extensions,” March 1999.).
| TOC |
Malware websites have proliferated recently, making malware and bot networks a major problem for users. In many cases, the initial contact with a virus or malware occurs when an unsuspecting user visits a particular website. This has even been observed to occur when a user visits an otherwise legitimate website, which contains external references that happen to contain malware, for example (such as advertisements served by a third party). Many organizations maintain lists of domains and FQDNs which host malware.
| TOC |
A user, malware agent, or bot requests a URL www.example.net or domain example.net. This site is associated with distributing malware or some other malicious activity that would not be desired by the user. The correct IP address is returned by the DNS and the user accesses the malware site or domain and their computer is infected with a bot.
| TOC |
By using Malicious Site Protection, a user may have their DNS response redirected from the IP address for the malicious URL www.example.net or domain example.net to a safe website that explains why the user was redirected. Importantly, the application attempting to access a malicious resource may or may not be a web browser and, further, may be operating without the user's knowledge and/or permission. This page on the aforementioned safe website that the user is directed to may also provide the user with a link to a method of opting out in the future. See Figure 3 (Malicious Domain Response) and Figure 9 (Malicious Site Redirect and HTTP Flow) for examples below. There may also be limited cases where it could be harmful to the objective of Malicious Site Protection to redirect the user to a safe website, in which case the user may not be directed to any resource, and a NXDOMAIN response be provided.
| TOC |
It is important to note that this technology can directly impact non-web clients such as instant messaging, VPNs, FTP, email filters-related DNS queries. Thus, special exclusions may need to be made in order to prevent unintentional side effects. Design considerations for the Web Error Search and Malicious Site Protection services should include properly and promptly terminating non-HTTP connection requests. A range of resource records may be redirected, such as A, AAAA, MX, SRV, and NAPTR records, in order to fulfill the objective of preventing access to certain network elements containing malicious content or which and in some way used to transmit, relay, or otherwise transfer malicious content. All other resource record types must be answered as if there was no redirection.
Malicious domain protection is also only effective if a user is actually using the DNS IP addresses that have this functionality. Thus, should a user's computer become compromised with some type of bot or virus that changes their DNS IP addresses (typically without their knowledge), the malicious domain protection would have no effect since the user is now pointed to DNS servers which are presumably in the control of a third party with malicious intentions.
| TOC |
ISPs and DNS ASPs MUST provide their users with a method to opt into (opt-in) or out (opt-out) of some or all DNS-based malware protection services. Opt-out and opt-in methods should be reliable and should take into consideration the Section 7 (Practices to Avoid) section below. Whether such services are offered on an opt-in or opt-out basis depends on a range of factors which are outside of the scope of this document. The two different methods, opt-out and opt-in, are described below.
| TOC |
Opt-Out is used when the users are by default offered all or some DNS-based malware protection services. As a result, the user must take an action to disable some or all such services. This is typically performed via a User Options Web Server. Users that have chosen to opt-out should receive DNS responses which are indistinguishable from those responses provided by a DNS server with no DNS Redirect functionality. In addition, opt-out should be persistent in nature, which means that opt-out should be tied to a fixed credential or attribute of some type, such as an account identifier, billing identifier, or equipment identifier, which is not typically subject to change on a regular basis.
| TOC |
Opt-In is used when the users are by default not offered any DNS-based malware protection services. As a result, the user must take an action to enable some or all such services. This is typically performed via a User Options Web Server.
| TOC |
Once a user has selected to opt-in or opt-out of DNS-based malware protection services, such changes should occur automatically, when this is technically possible, without requiring the user to manually change any settings on their computing device. Such changes should also occur within a reasonable period of time. In some cases, however, a user may be offered the ability to speed the period of time for these changes to take effect, such as by restarting the computing device or a piece of network equipment which connects them to their ISP's network, for example.
While an automated mechanism may be the easiest for users, since it requires no manual reconfiguration of their network settings, the authors also recognize that there may be extenuating circumstances where this is not achievable. In such cases, which may for example be due to the particular attributes of one or another ISP's network design, a fully automated mechanism may not be possible. Another example is where a user is switching from their ISP's DNS server IP addresses to those of a DNS ASP. As a result, a user in all of these cases, as well as other possible cases, must manually reconfigure their network with different DNS IP addresses.
| TOC |
There are several workable methods that can be employed to effect the actual opt-out for a given user. These include setting a local user application attribute, such as via a cookie in a web browser, as well as setting a network attribute, via a DHCP change or manually configuring the DNS IP addresses (in the operating system, modem, home gateway device, or router) in order to change the DNS IP addresses for a particular user.
While all of these methods are workable and can be made reliable, the best current method is via a network-based change of some sort. In this way, all Internet-connected computing devices within a given household are included in the opt-out (these devices are generally connected in some manner to the LAN side of some type of customer premise device, such as a cable modem or DSL modem). This is in contrast to a method which uses a local user application attribute, such as a cookie in a web browser, where deletion of cookies, upgrade to a new operating system, upgrade to a new web browser, use of a different web browser, or countless other factors on that device could cause the user to be opted back into a DNS-based malware protection service. Thus, a network-based approach which sets opt-out-related attributes at the device, or household level, is the most inclusive and persistent method for providing a reliable opt-out method, and is the recommended practice.
| TOC |
This document primarily focuses on the recommended practices for an ISP or ASP to provide users with DNS-based malware protection services. However, it is important to note that some entities may not operate in accordance with such practices. As such, some of these are catalogued below in order to contrast them with recommended practices and provide information which may be of interest and use to the community.
| TOC |
When DNSSEC has been implemented in a service provider's resolvers, DNS redirect MUST NOT be used, as it is technically incompatible with DNSSEC and breaks the chain of trust critical to proper DNSSEC validation functionality.
| TOC |
It has been observed that some service providers improperly utilize DNS-based malware protection services when there is a valid DNS resource record returned in response to a DNS recursive query. The effect is to redirect users to a server not maintained by the intended destination, such as a web site that looks like the intended web site but is not actually the intended site and is instead controlled by the service provider. For example a DNS query for www.example.com results in a valid A record response, but this valid response is instead replaced with an A record controlled by the service provider. In this case the intended server identified with the valid A record contained valid, lawful, non-malicious content, and there would otherwise appear to be no valid justification for a redirect to occur. See Figure 12 (Improper Redirect of Valid Response Redirect and HTTP Flow) for an example below.
If there is a valid and reasonable justification for such a redirect to occur, examples of which are not currently known by the authors of this document, then the resulting connection to the server that the user has been redirected to should clearly and prominently disclose that this is not the intended site. For example, in the case of an attempt by a user to connect to a web site, the site may contain a banner or frame which indicates that this is not the intended site or that the site is in some manner controlled by the service provider. In addition, such a notice should also offer a clear method to opt-out of this redirect function.
Thus, to summarize, redirection of valid responses SHOULD NOT be performed.
| TOC |
Redirection of SERVFAIL responses SHOULD NOT occur. SERVFAIL responses may occur intermittently in an operational network for a variety of highly transient reasons. As a result, a DNS Redirect should not be performed when a SERVFAIL response is received, as normal retry a short time later is likely to result in a valid response.
| TOC |
There are several well known and dependable methods of opt-out mechanisms that ISPs and DNS ASPs can deploy for users to opt-out of their DNS-based malware protection services. These methods can rather easily be employed and are highly recommended, as noted in Section 6 (Opt-In or Opt-Out Mechanisms). However, some ISPs and DNS ASPs may instead choose to employ a less dependable mechanism, which routinely fails to work as expected by users or is known not to function properly.
For example, one routinely unreliable method for opt-out is the cookie-based method. When a user opts out of a DNS-based malware protection service, a cookie is installed in their web browser. The problem with this method occurs when a user clears their cookies or the cookies are deleted for some reason. In some cases, users may configure their web browsers to clear all cookies every time the close their web browser. Thus, one possible effect upon the user in this case is that they are once again opted into the redirect service. Furthermore, a cookie-based method has the effect of only opting out browser-based protocols (generally HTTP and HTTPS), which means that the user may have non-web applications affected by DNS Redirect, even though they believe they have opted-out. As a result, there is no assured permanency with this opt-out method, nor does it work consistently across all applications and protocols, which can be aggravating to users who do not wish to utilize DNS-based malware protection services.
Another example of an unreliable method for opt-out is one where opt-out is tied to the IP address of the user, where that address may be subject to change on a regular basis, such as via an ISP-based DHCP lease. In such a case, if opt-out was tied to what can be considered a largely dynamic IP address, then the user would be opted-in every time they received a new IP address, forcing them to repeatedly opt-out.
Thus, to summarize, the opt-out mechanism provided to users SHOULD be reliable and SHOULD NOT be routinely broken, purposefully broken, or otherwise unreliable.
| TOC |
An ISP or DNS ASP should also understand that DNS query latency, the time between when a user's stub resolver issues and DNS query and receives a DNS response, should be kept as low as is reasonably possible. High DNS query latency is often perceived by users, and can have an adverse effect on a variety of applications where low DNS query latency may be especially important. Any additional processing which must be performed in order to provide DNS-based malware protection services should be monitored closely, in order that DNS Redirect functionality does not markedly slow DNS query performance.
Thus, to summarize, when DNS redirect is performed, DNS query performance SHOULD NOT suffer as a result, since this could provide an incrementally inferior user experience as compared to when DNS redirect is not performed.
| TOC |
Some users may decide to use the DNS server IP addresses of a DNS ASP or other non-ISP-provided DNS servers. Such selections should be preserved as the free choice of a user, particularly when DNS-based malware protection services are offered. Thus, an ISP SHOULD NOT redirect port 53 DNS traffic from servers intended by the user via their selection of non-ISP DNS servers to the DNS servers of the ISP, except in reasonable and justifiable cases where a user has been placed into a so-called "walled garden" for reasons of abuse, security compromise, account non-payment, new service activation, etc.
| TOC |
The functional design described in this section is intended to be generally representative of the many different ways that DNS-based malware protection services are deployed today. As such, they are necessarily high level and different implementations may vary somewhat, due to any number of factors.
| TOC |
The DNS Recursive Resolver is used by the host computer to translate fully qualified domain names into IP addresses, according to Section 3.6.1 of [RFC1034] (Mockapetris, P., “Domain names - concepts and facilities,” November 1987.). When a FQDN does not exist in authoritative DNS a NXDOMAIN response, as described in Section 4.1.1 of [RFC1035] (Mockapetris, P., “Domain names - implementation and specification,” November 1987.) is normally returned (see Figure 1 (DNS Redirect Response)). In the case of DNS Redirect, the NXDOMAIN response is changed to reply with a resource record (RR) response which instructs the host computer to send the original request to a new IP address (see Figure 1 (DNS Redirect Response)).
Request Request
www.example.invalid www.example.invalid
+--------+ +--------+
++--++ ------------> | | ------------> | |
|| || | | | |
+-++--++-+ | | | |
+--------+ <------------ | | <------------ | |
Host NXDOMAIN +--------+ NXDOMAIN +--------+
Computer Response Recursive Response Authoritative
Server Server
| Figure 1: DNS Redirect Response |
| TOC |
When a user requests an invalid URL or Domain, their web client is redirected to a Web Error Landing Server which presents several possible helpful website views (see Figure 2 (Web Error Landing Server)). The first is "Did you mean..." response which presents the user with possible correct results based on their original invalid request. The search server can also present search engine results to the user.
Request Request
www.example.invalid www.example.invalid
+--------+ +--------+
++--++ ---------------> | | --------------->| |
|| || | | | |
+-++--++-+ | | | |
+--------+ <-------------- | | <------------- | |
Host Redirect +--------+ NXDOMAIN +--------+
Computer IP Address Recursive Response Authoritative
Server Server
|
| ___________________________________
| +--------+ | Web Response: |
| | | | "Did you mean...www.example.com"|
+------> | | ------> |__________________________________|
| | | Search result: #1 |
| | | Search result: #2 |
+--------+ |__________________________________|
Web Server
Landing Page
| Figure 2: Web Error Landing Server |
| TOC |
The Web Browser Client is redirected to a Web Server Landing Page instead of presenting an error page when there is no valid DNS record present.
Examples of common Web Browser Clients include:
| TOC |
There may be certain domains which should be not be redirected under any circumstances for legal, business, or other reasons. The Domain White List can contain both domains, such as *.example.com, as well as specific FQDNs, such as www.example.com. For instance, the owner of example.com may request that the ISP or DNS ASP not perform DNS Redirect for the example.com domain, so that there is no DNS Redirect resulting from queries to nonexistent names, such as invalid.example.com.
| TOC |
Using a Malicious Domain List, a DNS server can redirect DNS requests that were intended for malicious websites or domains to a web server landing page (see Figure 3 (Malicious Domain Response)). The Malicious Domain List can contain both domains, such as *.example.net, as well as specific FQDNs, such as www.example.net.
Request Request
www.example.net www.example.net
+--------+ +--------+
++--++ ---------------> | | --------------->| |
|| || | | | |
+-++--++-+ | | | |
+--------+ <-------------- | | <------------- | |
Host Redirect +--------+ Response +--------+
Computer IP Address Recursive IP Address Authoritative
Server Server
|
| ___________________________________
| +--------+ | Web Response: |
| | | | "Malware software alert!" |
+-------> | | ------> |__________________________________|
| | | The site you attempted to access |
| | | is known to host malware that |
+--------+ | could damage your computer. |
Web Server |__________________________________|
Landing Page
| Figure 3: Malicious Domain Response |
| TOC |
Using a Malicious Domain List, a DNS server can redirect DNS requests that were intended for malicious websites or domains to a web server landing page (see Figure 3 (Malicious Domain Response)). The Legally-Mandated DNS Redirect Domain List can contain both domains, such as *.illegalcontent.example, as well as specific FQDNs, such as www.illegalcontent.example.net.
Request Request
www.illegalcontent.example.net www.illegalcontent.example.net
+--------+ +--------+
++--++ ---------------> | | --------------->| |
|| || | | | |
+-++--++-+ | | | |
+--------+ <-------------- | | <------------- | |
Host Redirect +--------+ Response +--------+
Computer IP Address Recursive IP Address Authoritative
Server Server
|
| ___________________________________
| +--------+ | Web Response: |
| | | | "Illegal content alert!" |
+-------> | | ------> |__________________________________|
| | | The site you attempted to access |
| | | has been designated as hosting |
+--------+ | illegal content under XYZ law of |
Web Server | 2009, which your ISP is legally |
Landing Page | compelled to block or be fined. |
|__________________________________|
| Figure 4: Legally-Mandated DNS Redirect Domain Response |
| TOC |
Using a Content Protection List, a DNS server can redirect DNS requests that were intended for websites or domains containing content deemed inappropriate by a user, to a web server landing page (see Figure 5 (Content-Based Redirect Domain Response)). The Legally-Mandated DNS Redirect Domain List can contain both domains, such as *.inappropriate.example, as well as specific FQDNs, such as www.inappropriate.example.com.
Request Request
www.inappropriate.example.com www.inappropriate.example.com
+--------+ +--------+
++--++ ---------------> | | --------------->| |
|| || | | | |
+-++--++-+ | | | |
+--------+ <-------------- | | <------------- | |
Host Redirect +--------+ Response +--------+
Computer IP Address Recursive IP Address Authoritative
Server Server
|
| ___________________________________
| +--------+ | Web Response: |
| | | | "Inappropriate content alert!" |
+------> | | ------> |__________________________________|
| | | The site you attempted to access |
| | | has been blocked by you or |
+--------+ | your network administrator. |
Web Server |__________________________________|
Landing Page
| Figure 5: Content-Based Redirect Domain Response |
| TOC |
| TOC |
This example represents a successful lookup of a valid DNS RR, and the resulting HTTP transaction.
Web DNS R DNS A DNS Web Server Browser Client Server Server 10.1.10.10 | Request | A | | | |www.example. |Record Query | A | | | com |www.example. |Record Query | | |------------>| com |www.example. | | | |------------>| com | | | | |------------>| | | | | A Record | | | | A Record | 10.1.10.10 | | | DNS Response| 10.1.10.10 |<------------| | | 10.1.10.10 |<------------| | | |<------------| | | | | HTTP GET | | | | | 10.1.10.10 | | | | |------------------------------------------------------>| | | | | | | | | | | | | | | |
| Figure 6: Successful DNS Lookup and HTTP Flow |
| TOC |
This example represents a lookup of a nonexistent DNS RR, and the resulting HTTP transaction.
Web DNS R DNS A DNS
Browser Client Server Server
| Request | A | |
|www.example. |Record Query | A |
| invalid |www.example. |Record Query |
|------------>| invalid |www.example. |
| |------------>| invalid |
| | |------------>|
| | | NXDOMAIN |
| | NXDOMAIN |<------------|
| NXDOMAIN |<------------| |
|<------------| | |
| | | |
| Figure 7: Unsuccessful DNS Lookup and HTTP Flow |
| TOC |
This example represents a lookup of a non-existing DNS RR, and the HTTP transition that results from a typical DNS-based malware protection service.
Redirect
Host R DNS A DNS Web Server Web Server
Computer Server Server 10.2.20.20 10.1.10.10
| A | | | |
|Record Query | A | | |
|www.example. |Record Query | | |
| invalid |www.example. | | |
|------------>| invalid | | |
| |------------>| | |
| A Record | NXDOMAIN | | |
| 10.2.20.20 |<------------| | |
|<------------| | | |
| HTTP GET | | | |
| 10.2.20.20 | | | |
|---------------------------------------->| |
| | | HTTP 200 OK | |
|<----------------------------------------| |
| A | | | |
|Record Query | A | | |
|www.example. |Record Query | | |
| com |www.example. | | |
|------------>| com | | |
| |------------>| | |
| | A Record | | |
| A Record | 10.1.10.10 | | |
| 10.1.10.10 |<------------| | |
|<------------| | | |
| HTTP GET | | | |
| 10.1.10.10 | | | |
|------------------------------------------------------>|
| | | | HTTP 200 OK |
|<------------------------------------------------------|
| | | | |
| Figure 8: DNS Redirect and HTTP Flow |
| TOC |
This example represents a lookup of a valid RR which hosts malware, and the HTTP transaction that results from a typical Malicious Site Protection service.
R DNS Mal Redirect
Host R DNS Server Web Server
Computer Server Mal List 10.2.20.20
| A | | |
|Record Query | Mal List | |
|www.example. | Query | |
| net |www.example. | |
|------------>| net | |
| |------------>| |
| | Postivie | |
| A Record | Match | |
| 10.2.20.20 |<------------| |
|<------------| | |
| HTTP GET | | |
| 10.2.20.20 | | |
|---------------------------------------->|
| | | HTTP 200 OK |
|<----------------------------------------|
| | | |
| Figure 9: Malicious Site Redirect and HTTP Flow |
| TOC |
This example represents a lookup of a valid RR which hosts illegal content, and the HTTP transaction that results from a typical Legally-Mandated DNS Redirect function.
R DNS Illegal Content
Server Redirect
Host R DNS Illegal Web Server
Computer Server Content List 10.2.20.20
| A | | |
|Record Query | Illegal | |
| www. | Content List | |
|illegalcontent.| | |
|example.net | Query | |
| | www. | |
| |illegalcontent.| |
| |example.net | |
|-------------->| | |
| |-------------->| |
| | Postivie | |
| A Record | Match | |
| 10.2.20.20 |<--------------| |
|<--------------| | |
| HTTP GET | | |
| 10.2.20.20 | | |
|---------------------------------------------->|
| | | HTTP 200 OK |
|<----------------------------------------------|
| | | |
| Figure 10: Legally-Mandated Redirect and HTTP Flow |
| TOC |
This example represents a lookup of a valid RR which hosts content which has been deemed inappropriate by a user, and the HTTP transaction that results from a typical Content-Based Redirect function.
R DNS Inappropriate
Server Content Redirect
Host R DNS Inappropriate Web Server
Computer Server Content List 10.2.20.20
| A | | |
|Record Query | Illegal | |
| www. | Content List | |
|inappropriate. | | |
|example.com | Query | |
| | www. | |
| |inappropriate. | |
| |example.com | |
|-------------->| | |
| |-------------->| |
| | Postivie | |
| A Record | Match | |
| 10.2.20.20 |<--------------| |
|<--------------| | |
| HTTP GET | | |
| 10.2.20.20 | | |
|---------------------------------------------->|
| | | HTTP 200 OK |
|<----------------------------------------------|
| | | |
| Figure 11: Content-Based Redirect and HTTP Flow |
| TOC |
This example represents an improper redirect occurring when a valid DNS RR should have been returned in response to a DNS recursive query for an example website, the resulting HTTP transaction, and that no DNS query or HTTP traffic was sent to the valid authoritative DNS server and valid web server. Section 10 (DNSSEC Considerations) below shows one of the reasons why this practice is problematic. Another reason is that a user intends to visit a valid resource with lawful and legitimate content, such as a web site, and is instead sent to a different destination (which may even closely resemble the intended site, in the pattern used by phishing sites).
R DNS Improper Valid
Server Redirect Valid Web
Host R DNS Improper Web Server A DNS Server
Computer Server Reirect List 10.2.20.20 Server 10.1.10.10
| A | Improper | | | |
|Record Query |Redirect List| | | |
|www.example. | Query | | | |
| com |www.example. | | | |
|------------>| com | | | |
| |------------>| | | |
| | Postivie | | | |
| A Record | Match | | | |
| 10.2.20.20 |<------------| | | |
|<------------| | | | |
| HTTP GET | | | | |
| 10.2.20.20 | | | | |
|-------------------------------------->| | |
| | |HTTP 200 OK| | |
|<--------------------------------------| | |
| | | | | |
| Figure 12: Improper Redirect of Valid Response Redirect and HTTP Flow |
| TOC |
DNS security extensions defined in [RFC4033] (Arends, R., Austein, R., Larson, M., Massey, D., and S. Rose, “DNS Security Introduction and Requirements,” March 2005.), [RFC4034] (Arends, R., Austein, R., Larson, M., Massey, D., and S. Rose, “Resource Records for the DNS Security Extensions,” March 2005.), and [RFC4035] (Arends, R., Austein, R., Larson, M., Massey, D., and S. Rose, “Protocol Modifications for the DNS Security Extensions,” March 2005.) use cryptographic digital signatures to provide origin authentication and integrity assurance for DNS data. This is done by creating signatures for DNS data on a Security-Aware Name Server that can be used by Security-Aware Resolvers to verify the answers. As the DNS redirections described herein take place on the recursive server there is no need to look into the communication between the recursive resolvers and name servers. Depending on the security awareness of recursive server used to perform DNS-based malware protection services, as well as the security awareness of the stub resolvers the following impact is observed on DNS Redirect when giving out answers to fully secured zones or parent zones:
So the only case where DNS security extensions cause problems for DNS Redirect is with a validating stub resolver. This case doesn't have widespread deployment now and could be mitigated by using trust anchor, configured by the applicable ISP or DNS ASP, that could be used to sign the redirected answers. As noted above in Section 9.7 (Improper Redirect of Valid Response Redirect and HTTP Flow), such improper redirection of valid responses may also cause DNSSEC trust verification problems.
| TOC |
Security best practices should be followed regarding access to the opt-in and opt-out functions, in order that someone other than the user is able to change the user's DNS Redirect settings. For example, the User Options Web Server must not permit someone to modify a page URI to access and change another user's options. Thus, if the URI is "http://www.example.net/redirect-options.php?account=1234", someone must not be able to modify the account to be "=1235" and then be able to change the options for a different user with some other additional validation being performed. While web site security practices are outside the scope of this document, the authors believe it is important to identify such problematic use cases to any ISPs and DNS ASPs offering and/or implementing DNS Redirect functionality.
| TOC |
There are no IANA considerations in this document.
| TOC |
The following people made significant textual contributions to this document and played an important role in the development and evolution of this document:
Don Bowman, Sandvine (don@sandvine.com)
Rick Hiester, Verizon (richard.hiester@verizon.com)
Chris Roosenraad, Time Warner Cable (chris.roosenraad@twcable.com)
David Ulevitch, OpenDNS (david@opendns.com)
| TOC |
The authors and contributors also wish to acknowledge the assistance of the following individuals in helping us to develop and/or review this document:
John Barnitz, Comcast Cable Communications (john_barnitz@cable.comcast.com)
Mike Burns, Cablevision (mburns@cablevision.com)
Phil Marcella, Comcast Interactive Media (phillip_marcella@cable.comcast.com)
Luis Uribarri, Comcast Cable Communications (luis_uribarri@cable.comcast.com)
Sandy Wilbourn, Nominum (sandy.wilbourn@nominum.com)
Matt Williams, Cox Cable (matt.williams@cox.com)
The authors and contributors also wish to thank ICANN's Security and Stability Advisory Committee (SSAC) for their review and debate of this document, as well as for raising important questions concerning DNSSEC compatibility.
| TOC |
| TOC |
[RFC Editor: This section is to be removed before publication]
-00: First version published
| TOC |
[RFC Editor: This section is to be removed before publication]
| TOC |
| Tom Creighton | |
| Comcast Cable Communications | |
| One Comcast Center | |
| 1701 John F. Kennedy Boulevard | |
| Philadelphia, PA 19103 | |
| US | |
| Email: | tom_creighton@cable.comcast.com |
| URI: | http://www.comcast.com |
| Chris Griffiths | |
| Comcast Cable Communications | |
| One Comcast Center | |
| 1701 John F. Kennedy Boulevard | |
| Philadelphia, PA 19103 | |
| US | |
| Email: | chris_griffiths@cable.comcast.com |
| URI: | http://www.comcast.com |
| Jason Livingood (editor) | |
| Comcast Cable Communications | |
| One Comcast Center | |
| 1701 John F. Kennedy Boulevard | |
| Philadelphia, PA 19103 | |
| US | |
| Email: | jason_livingood@cable.comcast.com |
| URI: | http://www.comcast.com |
| Ralf Weber | |
| Unaffiliated | |
| Bleichgarten 1 | |
| Hohenahr-Hohensolms 35644 | |
| Germany | |
| Email: | rw@hohensolms.de |