Google

bemasc@google.com

Akamai Technologies

mbishop@evequefou.be

Akamai Technologies

erik+ietf@nygren.org

General HTTP Working Group Internet-Draft This document specifies an “HTTPSSVC” DNS resource record type to facilitate the lookup of information needed to make connections for HTTPS URIs. The HTTPSSVC DNS RR mechanism allows an HTTPS origin hostname to be served from multiple network services, each with associated parameters (such as transport protocol and keying material for encrypting TLS SNI). It also provides a solution for the inability of the DNS to allow a CNAME to be placed at the apex of a domain name. Finally, it provides a way to indicate that the origin supports HTTPS without having to resort to redirects, allowing clients to remove HTTP from the bootstrapping process. By allowing this information to be bootstrapped in the DNS, it allows for clients to learn of alternative services before their first contact with the origin. This arrangement offers potential benefits to both performance and privacy. TO BE REMOVED: This proposal is inspired by and based on recent DNS usage proposals such as ALTSVC, ANAME, and ESNIKEYS (as well as long standing desires to have SRV or a functional equivalent implemented for HTTP). These proposals each provide an important function but are potentially incompatible with each other, such as when an origin is load-balanced across multiple hosting providers (multi-CDN). Furthermore, these each add potential cases for adding additional record lookups in-addition to AAAA/A lookups. This design attempts to provide a unified framework that encompasses the key functionality of these proposals, as well as providing some extensibility for addressing similar future challenges.

The HTTPSSVC RR is intended to address a number of challenges facing HTTPS clients and services, while also providing an extensible model to handle similar use-cases without forcing clients to perform additional DNS lookups and without forcing them to first make connections to a default service for the origin. When clients need to make a connection to fetch resources associated with an HTTPS URI, they must first resolve A and/or AAAA address resource records for the origin hostname. This is adequate when clients default to TCP port 443, do not support Encrypted SNI , and where the origin service operator does not have a desire to put an CNAME at a zone apex (such as for “https://example.com”). Handling situations beyond this within the DNS requires learning additional information, and it is highly desirable to minimize the number of round-trip and lookups required to learn this additional information.

As an introductory example, a set of example HTTPSSVC and associated A+AAAA records might be:

In the preceding example, both of the “example.com” and “www.example.com” origin names are aliased to use service endpoints offered as “svc.example.net” (with “www.example.com” continuing to use a CNAME alias). HTTP/2 is available on a cluster of machines located at svc2.example.net with TCP port 8002 and HTTP/3 is available on a cluster of machines located at svc3.example.net with UDP port 8003. An ESNI key is specified which allows the SNI values of “example.com” and “www.example.com” to be encrypted in the handshake with these service endpoints. When connecting, clients will continue to treat the authoritative origins as “https://example.com” and “https://www.example.com”, respectively.

The goal of the HTTSSVC RR is to allow clients to resolve a single additional DNS RR in a way that: Provides service endpoints authoritative for an origin, along with parameters associated with each of these endpoints. In particular: to support connecting directly to (QUIC transport) service endpoints to obtain the keys associated with a service endpoint Does not assume that all service endpoints have the same parameters (such as ESNI keys) or capabilities (such as ) or are even operated by the same entity. This is important as DNS does not provide any way to tie together multiple RRs for the same name. For example, if www.example.com is a CNAME alias that switches between one of three CDNs or hosting enviroments, records (such as A and AAAA) for that name may have been sourced from different environments. Enables the functional equivalent of a CNAME at a zone apex (such as “example.com”) for HTTPS traffic, and generally enables delegation of operational authority for an HTTPS origin within the DNS to an alternate name. This addresses a set of long-standing issues due to HTTP(S) clients not implementing support for SRV records, as well as due to a limitation that a DNS name can not have both a CNAME record as well as NS RRs (as is the case for zone apex names)

This subsection briefly describes the HTTPSVC RR in a non-normative manner. The HTTPSSVC RR has four primary fields: SvcRecordType: A numeric flag indicating how to interpret the subsequent fields. When “0”, it indicates that the RR contains an alias. When “1”, it indicates that the RR contains an alternative service definition. SvcFieldPriority: The priority of this record (relative to others, with lower values preferred). Applicable when SvcRecordType is “1”, and otherwise has value “0”. (Described in .) SvcDomainName: The domain name of either the alias target (when SvcRecordType is “0”) or the uri-host domain name of the alternative service endpoint (when SvcRecordType is “1”). SvcFieldValue: An Alternative Service field value describing the alternative service endpoint for the domain name specified in SvcDomainName (only when SvcRecordType is “1” and otherwise empty). Cooperating DNS recursive resolvers will perform subsequent record resolution (for HTTPSSVC, A, and AAAA records) and return them in the Additional Section of the response. Clients must either use responses included in the additional section returned by the recursive resolver or perform necessary HTTPSSVC, A, and AAAA record resolutions. DNS authoritative servers may attach in-bailiwick HTTPSVC, A, AAAA, and CNAME records in the Additional Section to responses for an HTTPSVC query. When SvcRecordType is “1”, the HTTPSSVC RR extends the concept introduced in the HTTP Alternative Services proposed standard . Alt-Svc defines: an extensible data model for describing alternative network endpoints that are authoritative for an origin the “Alt-Svc Field Value”, a text format for representing this information standards for sending information in this format from a server to a client over HTTP/1.1 and HTTP/2. Together, these components provide a toolkit that has proven useful and effective for informing a client of alternative services for an origin. However, making use of an alternative service requires contacting the origin server first. This creates an obvious performance cost: users wait for a full HTTP connection initiation (multiple roundtrips) before learning of an alternative service that is preferred by the origin. The first connection also publicly reveals the user’s intended destination to all entities along the network path. The SvcFieldValue includes the Alt-Svc Field Value through the DNS. This is in its standard text format, with the uri-host portion of the alt-authority component moved into the SvcDomainName field of the HTTPSSVC RR. A client receiving this information during DNS resolution can skip the initial connection and proceed directly to an alternative service.

This document also defines one additional Alt-Svc parameter that can be used within SvcFieldValue: esnikeys (): The ESNIKeys structure from Section 4.1 of for use in encrypting the actual origin hostname in the TLS handshake.

For consistency with , we adopt the following definitions: An “origin” is an information source as in . The “origin server” is the server that the client would reach when accessing the origin in the absence of Alt-Svc. An “alternative service” is a different server that can serve the origin. Abstractly, the origin consists of a scheme (typically “https”), a host name, and a port (typically “443”). Additional DNS terminology intends to be consistent with . The key words “MUST”, “MUST NOT”, “REQUIRED”, “SHALL”, “SHALL NOT”, “SHOULD”, “SHOULD NOT”, “RECOMMENDED”, “NOT RECOMMENDED”, “MAY”, and “OPTIONAL” in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here.

The HTTPSSVC DNS resource record (RR) type (RRTYPE ???) is used to locate endpoints that can service an “https” origin. The presentation format of the record is: RRName TTL Class HTTPSSVC SvcRecordType SvcFieldPriority \ SvcDomainName SvcFieldValue where SvcRecordType is a numeric value of either “0” or “1”, SvcFieldPriority is a number in the range 0-65535, SvcDomainName is a domain name, and SvcFieldValue is a string present when SvcRecordType is “1”. The algorithm for resolving HTTPSSVC records and associated address records is specified in .

The RDATA for the HTTPSSVC RR consists of: a 1 octet flag field for SvcRecordType, interpreted as an unsigned numeric value (0 to 255, with only values “0” and “1” defined here) a 2 octet field for SvcFieldPriority as an integer in network byte order. If SvcRecordType is “0”, SvcFieldPriority MUST be 0. a 1 octet length field for the SvcDomainName. If SvcRecordType is “1”, a length of 0 indicates that uri-host is omitted. Otherwise, this value MUST NOT be 0. the uncompressed SvcDomainName of the specified length, represented as a sequence of length-prefixed labels as in Section 3.1 of . a 2 octet length field for the SvcFieldValue the SvcFieldValue byte string of the specified length (up to 65536 characters) When SvcRecordType is “0”, the length of SvcFieldValue SHOULD be 0 and clients MUST ignore the contents of non-empty SvcFieldValue fields.

In the case of the HTTPSSVC RR, an origin is translated into the RRName in the following manner: If the scheme is “https” and the port is 443, then the RRName is equal to the origin host name. Otherwise the RRName is represented by prefixing the port and scheme with “_”, then concatenating them with the host name, resulting in a domain name like “_832._https.www.example.com.”. When a prior CNAME or HTTPSSVC record has aliased to an HTTPSSVC record, RRName shall be the name of the alias target. Note that none of these forms alter the HTTPS origin or authority. For example, clients MUST continue to validate TLS certificate hostnames based on the origin host.

The SvcRecordType field is a numeric value defined to be either “0” or “1”. Within an HTTPSSVC RRSet, all RRs must have the same value for SvcRecordType. Clients and recursive servers MUST ignore HTTPSSVC resource records with other SvcRecordType values. If an RRSet contains a record with type “0”, the client MUST ignore any records in the set with type “1”. When SvcRecordType is “0”, the HTTPSSVC is defined to be in “alias form”. When SvcRecordType is “1”, the HTTPSSVC is defined to be in “alternative service form”.

When SvcRecordType is “0”, the HTTPSSVC record is to be treated similar to a CNAME alias pointing to the domain name specified in SvcDomainName. HTTPSSVC RRSets MUST only have a single resource record in this form. If multiple are present, clients or recursive resolvers SHOULD pick one non-determinstically. The common use-case for this form of the HTTPSVC record is as an alternative to CNAMEs at the zone apex where they are not allowed. For example, if an operator of https://example.com wanted to point HTTPS requests to a service operating at svc.example.net, they would publish a record such as:

The SvcDomainName MUST point to a domain name that contains another HTTPSSVC record and/or address (AAAA and/or A) records. Note that the RRName and the SvcDomainName MAY themselves be CNAMEs. Clients and recursive resolvers MUST follow CNAMEs as normal. Due to the risk of loops, clients and recursive resolvers MUST implement loop detection. Chains of consecutive HTTPSSVC and CNAME records SHOULD be limited to (8?) prior to reaching terminal address records. The SvcFieldValue in this form SHOULD be an empty string and clients MUST ignore its contents. As legacy clients will not know to use this record, service operators will likely need to retain fallback AAAA and A records alongside this HTTPSSVC record, although in a common case the target of the HTTPSSVC record might have better performance, and therefore would be preferable for clients implementing this specification to use.

When SvcRecordType is “1”, the combination of SvcDomainName and SvcFieldValue within each resource record associates an Alternative Service Field Value with an origin. The SvcFieldValue of the HTTPSSVC resource record contains an Alt-Svc Field Value, exactly as defined in Section 4 of , but with the uri-host moved to the SvcDomainName field. For example, if the operator of https://www.example.com intends to include an HTTP response header like

they could also publish an HTTPSSVC DNS RRSet like

This data type can be represented as an Unknown RR as described in :

This construction is intended to be extensible in two ways. First, any extensions that are made to the Alt-Svc format for transmission over HTTPS are also applicable here, unless expressly mentioned otherwise. Second, by defining a way to map non-HTTPS schemes and non-default ports (), we provide a way for the HTTPSSVC to be used for them as needed. However, by using the origin name for the RRName for scheme https and port 443 we allow HTTPSSVC records to be included at the end of CNAME chains for existing site implementations without requiring changes in the zone containing the origin.

Publishing an alternative services form HTTPSSVC record in DNS is intended to be equivalent to transmitting this field value over HTTPS, and receiving an HTTPSSVC record is intended to be equivalent to receiving this field value over HTTPS. However, there are some small differences in the intended client and server behavior.

When publishing an HTTPSSVC record in DNS, server operators MUST omit the “ma” parameter, which encodes the “max age” (i.e. expiration time) of an Alt-Svc Field Value. Instead, server operators SHOULD encode the expiration time in the DNS TTL, and MUST NOT set a TTL longer than the intended “max age”. When receiving an HTTPSSVC record, clients SHOULD synthesize a new “ma” parameter from the DNS TTL if the resulting alt-value is being passed to a subsystem that might employ caching. When publishing an HTTPSSVC record, server operators MUST omit the “persist” parameter, which indicates whether the client should use this record on other network paths. When receiving an HTTPSSVC record, clients MUST discard any records that contain a “persist” flag. Disabling persistence is important to prevent a local adversary in one network from implanting a forged DNS record that allows them to track users or hinder their connections after they leave that network.

Server operators MAY publish multiple SvcRecordType “1” HTTPSSVC records as an RRSET. When converting a collection of alt-values into an HTTPSSVC RRSET, the server operator MUST set the overall TTL to a value no larger than the minimum of the “max age” values (following Section 5.2 of ). Each RR MUST contain exactly one alt-value, as described in Section 3 of . As RRs within an RRSET are explicitly unordered collections, the SvcFieldPriority value is introduced to indicate priority. HTTPSSVC RRs with a smaller SvcFieldPriority value SHOULD be given preference over RRs with a larger SvcFieldPriority value. Alt-values received via HTTPS are preferred over any Alt-value received via DNS. When receiving an RRSET containing multiple HTTPSSVC records with the same SvcFieldPriority value, clients SHOULD apply a random shuffle within a priority level to the records before using them, to ensure randomized load-balancing.

For a client to construct the equivalent of an Alt-Svc HTTP response header: For each RR, the SvcDomainName MUST be inserted as the uri-host. The RRs SHOULD be ordered by increasing SvcFieldPriority, with shuffling for equal SvcFieldPriority values. Clients MAY choose to further prioritize alt-values where address records are immediately available for the alt-value’s SvcDomainName. The client SHOULD concatenate the thus-transformed-and-ordered SvcFieldValues in the RRSET, separated by commas. (This is semantically equivalent to receiving multiple Alt-Svc HTTP response headers, according to Section 3.2.2 of ).

Sending Alt-Svc over HTTP allows the server to tailor the Alt-Svc Field Value specifically to the client. When using an HTTPSSVC DNS record, groups of clients will necessarily receive the same Alt-Svc Field Value. Therefore, this standard is not suitable for uses that require single-client granularity in Alt-Svc. Some DNS caching systems incorrectly extend the lifetime of DNS records beyond the stated TTL. Server operators MUST NOT rely on HTTPSSVC records expiring on time, and MAY shorten the TTL to compensate.

When attempting to resolve a name HOST, clients should follow in-order: Issue parallel AAAA/A and HTTPSSVC queries for the name HOST. The answers for these may or may not include CNAME pointers before reaching one or more of these records. If an HTTPSSVC record of SvcRecordType “0” is returned for HOST, clients should loop back to step 1 replacing HOST with SvcDomainName, subject to loop detection heuristics. If one or more HTTPSSVC record of SvcRecordType “1” is returned for HOST, clients should synthesize equivalent Alt-Svc Field Values based on the SvcDomainName and SvcFieldValue. If one of these alt-values is selected to be used in a connection, the client will need to resolve AAAA and/or A records for SvcDomainName. If only AAAA and/or A records are present for HOST (and no HTTPSSVC), clients should make a connection to one of the IP addresses contained in these records and proceed normally. When selecting between AAAA and A records to use, clients may use an approach such as Some possible optimizations are discussed in to reduce latency impact in comparison to ordinary AAAA/A lookups.

By publishing an HTTPSSVC record, the server operator indicates that all useful HTTP resources on that origin are reachable over HTTPS, similar to HTTP Strict Transport Security . When an HTTPSSVC record is present for an origin, all “http” scheme requests for that origin SHOULD logically be redirected to “https”. Prior to making an “http” scheme request, the client SHOULD perform a lookup to determine if an HTTPSSVC record is available for that origin. To do so, the client SHOULD construct a corresponding “https” URL as follows: Replace the “http” scheme with “https”. If the “http” URL explicitly specifies port 80, specify port 443. Do not alter any other aspect of the URL. This construction is equivalent to Section 8.3 of , point 5. If an HTTPSSVC record is present for this “https” URL, the client should treat this as the equivalent of receiving an HTTP “302 Found” redirect to the “https” URL. Because HTTPSSVC is received over an often insecure channel (DNS), clients MUST NOT place any more trust in this signal than if they had received a 302 redirect over cleartext HTTP. If the HTTPSSVC query results in a SERVFAIL error, and the connection between the client and the recursive resolver is cryptographically protected (e.g. using TLS or HTTPS ), the client SHOULD abandon the connection attempt and display an error message. A SERVFAIL error can occur if the domain is DNSSEC-signed, the recursive resolver is DNSSEC-validating, and an active attacker between the recursive resolver and the authoritative DNS server is attempting to prevent the upgrade to HTTPS. Similarly, if the client enforces DNSSEC validation on A/AAAA RRs, it SHOULD abandon the connection attempt if the HTTPSSVC RR fails to validate.

If the client has an Alt-Svc cache, and a usable Alt-Svc value is present in that cache, then the client SHOULD NOT issue an HTTPSSVC DNS query. Instead, the client SHOULD proceed with alternative service connection as usual. If the client has a cached Alt-Svc entry that is expiring, the client MAY perform an HTTPSSVC query to refresh the entry.

Recursive DNS servers SHOULD resolve SvcDomainName records and include them in the Additional Section (along with any relevant CNAME records). For SvcRecordType=0, recursive DNS servers SHOULD attempt to resolve and include A, AAAA, and HTTPSSVC records. For SvcRecordType=1, recursive DNS servers SHOULD attempt to resolve and include A and AAAA records. Authoritative DNS servers SHOULD return A, AAAA, and HTTPSSVC records (as well as any relevant CNAME records) in the Additional Section for any in-bailiwick SvcDomainNames.

For optimal performance (i.e. minimum connection setup time), clients SHOULD issue address (AAAA and/or A) and HTTPSSVC queries simultaneously, and SHOULD implement a client-side DNS cache. With these optimizations in place, and conforming DNS servers, using HTTPSSVC does not add network latency to connection setup. A nonconforming recursive resolver might return an HTTPSSVC response with a nonempty SvcDomainName, without the corresponding address records. If all the HTTPSSVC RRs in the response have nonempty SvcDomainName values, and the client does not have address records for any of these values in its DNS cache, the client SHOULD perform an additional address query for the selected SvcDomainName. The additional DNS query in this case introduces a delay. To avoid causing a delay for clients using a nonconforming recursive resolver, domain owners SHOULD choose the SvcDomainName to be a name in the origin hostname’s CNAME chain if possible. This will ensure that the required address records are already present in the client’s DNS cache as part of the responses to the address queries that were issued in parallel. Highly performance-sensitive clients MAY implement the following special- case shortcut to avoid increased connection time: if (1) one of the HTTPSSVC records returned has SvcRecordType=0, (2) its SvcDomainName is not in the DNS cache, and (3) the address queries for the origin domain return usable IP addresses, then the client MAY ignore the HTTPSSVC records and connect directly to the origin domain. When the SvcDomainNames and any needed HTTPSSVC records are available, the client SHOULD make subsequent requests over connections specified by the HTTPSSVC records. Server operators can therefore expect that publishing HTTPSSVC records with SvcRecordType=0 should not cause an additional DNS query for performance-sensitive clients. Server operators who wish to prevent this optimization should use SvcRecordType=1.

An Alt-Svc “esnikeys” parameter is defined for specifying ESNI keys corresponding to an alternative service. The value of the parameter is an ESNIKeys structure encoded in , or the empty string. ESNI-aware clients SHOULD prefer alt-values with nonempty esnikeys. This parameter MAY also be sent in Alt-Svc HTTP response headers and HTTP/2 ALTSVC frames. The Alt-Svc specification states that “the client MAY fall back to using the origin” in case of connection failure . This behavior is not suitable for ESNI, because fallback would negate the privacy benefits of ESNI. Accordingly, any connection attempt that uses ESNI MUST fall back only to another alt-value that also has the esnikeys parameter. If the parameter’s value is the empty string, the client SHOULD connect as it would in the absence of any ESNIKeys information. For example, suppose a server operator has two alternatives. Alternative A is reliably accessible but does not support ESNI. Alternative B supports ESNI but is not reliably accessible. The server operator could include a full esnikeys value in Alternative B, and mark Alternative A with esnikeys=”” to indicate that fallback from B to A is allowed.

The purpose of this standard is to reduce connection latency and improve user privacy. Server operators implementing this standard SHOULD also implement TLS 1.3 and OCSP Stapling , both of which confer substantial performance and privacy benefits when used in combination with HTTPSSVC records. To realize the greatest privacy benefits, this proposal is intended for use with a privacy-preserving DNS transport (like DNS over TLS or DNS over HTTPS ). However, performance improvements, and some modest privacy improvements, are possible without the use of those standards. This RRType could be extended to support schemes other than “https”. Any such scheme MUST have an entry under the HTTPSSVC RRType in the IANA DNS Underscore Global Scoped Entry Registry The scheme SHOULD have an entry in the IANA URI Schemes Registry . The scheme SHOULD be one for which Alt-Svc is defined.

Alt-Svc Field Values are intended for distribution over untrusted channels, and clients are REQUIRED to verify that the alternative service is authoritative for the origin (Section 2.1 of ). Therefore, DNSSEC signing and validation are OPTIONAL for publishing and using HTTPSSVC records. TBD: expand this section in more detail. In particular: * Just as with , clients must validate the TLS server certificate against hostname associated with the origin. Clients MUST NOT use the SvcDomainName as any part of the server TLS certificate validation. * …

Per , please add the following entry to the data type range of the Resource Record (RR) TYPEs registry: TYPE Meaning Reference HTTPSSVC HTTPS Service Location (This document) Per , please add the following entries to the DNS Underscore Global Scoped Entry Registry: RR TYPE _NODE NAME Meaning Reference HTTPSSVC _https Alt-Svc for HTTPS (This document) Per , please add the following entries to the HTTP Alt-Svc Parameter Registry: Alt-Svc Parameter Meaning Reference esnikeys Encrypted SNI keys (This document)

There have been a wide range of proposed solutions over the years to the “CNAME at the Zone Apex” challenge proposed. These include , , and others. Thank you to Ian Swett, Ralf Weber, Jon Reed, and others for their feedback and suggestions on this draft.

This document defines a simple mechanism for encrypting the Server Name Indication for TLS 1.3. The QUIC transport protocol has several features that are desirable in a transport for HTTP, such as stream multiplexing, per-stream flow control, and low-latency connection establishment. This document describes a mapping of HTTP semantics over QUIC. This document also identifies HTTP/2 features that are subsumed by QUIC, and describes how HTTP/2 extensions can be ported to HTTP/3. This document specifies "Alternative Services" for HTTP, which allow an origin's resources to be authoritatively available at a separate network location, possibly accessed with a different protocol configuration. This document defines the concept of an "origin", which is often used as the scope of authority or privilege by user agents. Typically, user agents isolate content retrieved from different origins to prevent malicious web site operators from interfering with the operation of benign web sites. In addition to outlining the principles that underlie the concept of origin, this document details how to determine the origin of a URI and how to serialize an origin into a string. It also defines an HTTP header field, named "Origin", that indicates which origins are associated with an HTTP request. [STANDARDS-TRACK] In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings. This RFC is the revised specification of the protocol and format used in the implementation of the Domain Name System. It obsoletes RFC-883. This memo documents the details of the domain name client - server communication. Extending the Domain Name System (DNS) with new Resource Record (RR) types currently requires changes to name server software. This document specifies the changes necessary to allow future DNS implementations to handle new RR types transparently. [STANDARDS-TRACK] This document considers some areas that have been identified as problems with the specification of the Domain Name System, and proposes remedies for the defects identified. [STANDARDS-TRACK] Many communication protocols operating over the modern Internet use hostnames. These often resolve to multiple IP addresses, each of which may have different performance and connectivity characteristics. Since specific addresses or address families (IPv4 or IPv6) may be blocked, broken, or sub-optimal on a network, clients that attempt multiple connections in parallel have a chance of establishing a connection more quickly. This document specifies requirements for algorithms that reduce this user-visible delay and provides an example algorithm, referred to as "Happy Eyeballs". This document obsoletes the original algorithm description in RFC 6555. This specification defines a mechanism enabling web sites to declare themselves accessible only via secure connections and/or for users to be able to direct their user agent(s) to interact with given sites only over secure connections. This overall policy is referred to as HTTP Strict Transport Security (HSTS). The policy is declared by web sites via the Strict-Transport-Security HTTP response header field and/or by other means, such as user agent configuration, for example. [STANDARDS-TRACK] This document describes the use of Transport Layer Security (TLS) to provide privacy for DNS. Encryption provided by TLS eliminates opportunities for eavesdropping and on-path tampering with DNS queries in the network, such as discussed in RFC 7626. In addition, this document specifies two usage profiles for DNS over TLS and provides advice on performance considerations to minimize overhead from using TCP and TLS with DNS.This document focuses on securing stub-to-recursive traffic, as per the charter of the DPRIVE Working Group. It does not prevent future applications of the protocol to recursive-to-authoritative traffic. This document defines a protocol for sending DNS queries and getting DNS responses over HTTPS. Each DNS query-response pair is mapped into an HTTP exchange. This document describes the commonly used base 64, base 32, and base 16 encoding schemes. It also discusses the use of line-feeds in encoded data, use of padding in encoded data, use of non-alphabet characters in encoded data, use of different encoding alphabets, and canonical encodings. [STANDARDS-TRACK] This document specifies version 1.3 of the Transport Layer Security (TLS) protocol. TLS allows client/server applications to communicate over the Internet in a way that is designed to prevent eavesdropping, tampering, and message forgery. This document updates RFCs 4492, 5705, and 6066 and it obsoletes RFCs 5077, 5246, and 6961. This document also specifies new requirements for TLS 1.2 implementations. This document provides specifications for existing TLS extensions. It is a companion document for RFC 5246, "The Transport Layer Security (TLS) Protocol Version 1.2". The extensions specified are server_name, max_fragment_length, client_certificate_url, trusted_ca_keys, truncated_hmac, and status_request. [STANDARDS-TRACK] Formally, any DNS resource record may occur under any domain name. However some services use an operational convention for defining specific interpretations of an RRset, by locating the records in a DNS branch, under the parent domain to which the RRset actually applies. The top of this subordinate branch is defined by a naming convention that uses a reserved node name, which begins with an _underscore. The underscored naming construct defines a semantic scope for DNS record types that are associated with the parent domain, above the underscored branch. This specification explores the nature of this DNS usage and defines the "DNS Global Underscore Scoped Entry Registry" with IANA. The purpose of the Underscore registry is to avoid collisions resulting from the use of the same underscore-based name, for different services. This document updates the guidelines and recommendations, as well as the IANA registration processes, for the definition of Uniform Resource Identifier (URI) schemes. It obsoletes RFC 4395. HTTP Alternative Services provides a mechanism for an origin to declare that its content is accessible via some other combination of host, port, and protocol. In the process of using such an alternative, an observer can identify that the client is requesting resources from a particular hostname. This document extends HTTP Alternative Services, in combination with Secondary Certificate Authentication, to enable clients not to disclose the origin to which they intend to connect. The Domain Name System (DNS) is defined in literally dozens of different RFCs. The terminology used by implementers and developers of DNS protocols, and by operators of DNS systems, has sometimes changed in the decades since the DNS was first defined. This document gives current definitions for many of the terms used in the DNS in a single document.This document obsoletes RFC 7719 and updates RFC 2308. The Hypertext Transfer Protocol (HTTP) is a stateless application-level protocol for distributed, collaborative, hypertext information systems. This document provides an overview of HTTP architecture and its associated terminology, defines the "http" and "https" Uniform Resource Identifier (URI) schemes, defines the HTTP/1.1 message syntax and parsing requirements, and describes related security concerns for implementations. This document specifies Internet Assigned Numbers Authority (IANA) parameter assignment considerations for the allocation of Domain Name System (DNS) resource record types, CLASSes, operation codes, error codes, DNS protocol message header bits, and AFSDB resource record subtypes. It obsoletes RFC 6195 and updates RFCs 1183, 2845, 2930, and 3597. This document specifies an "HTTP" resource record type for the DNS to facilitate the lookup of the server hostname of HTTP(s) URIs. It is intended to replace the use of CNAME records for this purpose, and in the process provides a solution for the inability of the DNS to allow a CNAME to be placed at the apex of a domain name. This document defines the "ANAME" DNS RR type, to provide similar functionality to CNAME, but only for type A and AAAA queries. Unlike CNAME, an ANAME can coexist with other record types. The ANAME RR allows zone owners to make an apex domain name into an alias in a standards compliant manner. This document describes a DNS RR which specifies the location of the server(s) for a specific protocol and domain. [STANDARDS-TRACK]

The HTTPSSVC record type closely resembles some existing record types and proposals. A complaint with all of the alternatives is that web clients have seemed unenthusiastic about implementing them. The hope here is that by providing an extensible solution that solves multiple problems we will overcome the inertia and have a path to achieve client implementation.

An SRV record can perform a similar function to the HTTPSSVC record, informing a client to look in a different location for a service. However, there are several differences: SRV records are typically mandatory, whereas clients will always continue to function correctly without making use of Alt-Svc or HTTPSSVC. SRV records cannot instruct the client to switch or upgrade protocols, whereas Alt-Svc can signal such an upgrade (e.g. to HTTP/2). SRV records are not extensible, whereas Alt-Svc and thus HTTPSSVC can be extended with new parameters. For example, this is what allows the incorporation of ESNI keys in HTTPSSVC. Using SRV records would not allow a client to skip processing of the Alt-Svc information in a subsequent connection, so it does not confer a performance advantage.

Unlike , this approach is extensible to cover Alt-Svc and ESNIKeys use-cases. Like that proposal, this addresses the zone apex CNAME challenge. Like that proposal it remains necessary to continue to include address records at the zone apex for legacy clients.

Unlike , this approach is extensible to cover Alt-Svc and ESNIKeys use-cases. This approach also does not require any changes or special handling on either authoritative or master servers, beyond optionally returning in-bailiwick additional records. Like that proposal, this addresses the zone apex CNAME challenge for clients that implement this. However with this HTTPSSVC proposal it remains necessary to continue to include address records at the zone apex for legacy clients. If deployment of this standard is successful, the number of legacy clients will fall over time. As the number of legacy clients declines, the operational effort required to serve these users without the benefit of HTTPSSVC indirection should fall. Server operators can easily observe how much traffic reaches this legacy endpoint, and may remove the apex’s address records if the observed legacy traffic has fallen to negligible levels.

Unlike , this approach is extensible and covers the Alt-Svc case as well as addresses the zone apex CNAME challenge. By using the Alt-Svc model we also provide a way to solve the ESNI multi-CDN challenges in a general case. Unlike ESNI, this is focused on the specific case of HTTPS, although this approach could be extended for other protocols. It also allows specifying ESNI keys for a specific port, not an entire host.

Defining an Alt-Svc sni= parameter (such as from ) would have provided some benefits to clients and servers not implementing ESNI, such as for specifying that “_wildcard.example.com” could be sent as an SNI value rather than the full name. There is nothing precluding HTTPSSVC from being used with an sni= parameter if one were to be defined, but it is not included here to reduce scope, complexity, and additional potential security and tracking risks.

This draft is intended to be a work-in-progress for discussion. Many details are expected to change with subsequent refinement. Some known issues or topics for discussion are listed below.

Naming is hard. The “HTTPSSVC” is proposed as a placeholder. Other names for this record might include ALTSVC, HTTPS, HTTPSSRV, B, or something else.

The focus of this record is on optimizing the common case of the “https” scheme. It is worth discussing whether this is a valid assumption or if a more general solution is applicable. Past efforts to over-generalize have not met with broad success.

The length of SvcFieldValue is redundant. Should we eliminate this? Advice from experts in DNS wire format best practices would be greatly appreciated to refine the proposed details, overall.

Only values of “0” and “1” are allowed for SvcRecordType. Should we give more thought to potential future values? The current version tries to leave this open by indicating that resource records with unknown SvcRecordType values should be ignored (and perhaps should be switched to MUST be ignored)?

The SvcFieldPriority could alternately be included as a pri= Alt-Svc attribute. It wouldn’t be applicable for Alt-Svc returned via HTTP, but it is also not necessarily needed by DNS servers. It is also not used when SvcRecordType=0. A related question is whether to omit it from the textual representation when SvcRecordType=0. Regardless, having a series of sequential numeric values in the textual representation has risk of user error, especially as MX, SRV, and others all have their own variations here.

Some other similar mechanisms such as SRV have a weight in-addition to priority. That is excluded here for simplicity. It could always be added as an optional Alt-Svc attribute.