Internet Engineering Task Force M. Blanchet Internet-Draft Viagenie Intended status: Informational W. Hardaker Expires: 19 May 2027 Google LLC 15 November 2026 Deployment and Use of the Domain Name System(DNS) in Deep Space draft-many-tiptop-dns-04 Abstract Deep space communications involve long delays (e.g., Earth to Mars has one-way delays 4-24 minutes) and intermittent communications, mainly because of orbital dynamics. This document lists operational methods to enable local DNS name resolving on celestial body networks such that there are no real-time query and response flow to name servers on (Earth) Internet. Status of This Memo 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 https://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 19 May 2027. Copyright Notice Copyright (c) 2026 IETF Trust and the persons identified as the document authors. All rights reserved. Blanchet & Hardaker Expires 19 May 2027 [Page 1] Internet-Draft DNS in Deep Space November 2026 This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://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 Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 2. Use Case and Requirements . . . . . . . . . . . . . . . . . . 3 3. Possible Approaches . . . . . . . . . . . . . . . . . . . . . 4 3.1. Pre-fetching of needed names and records . . . . . . . . 5 3.2. Pre-fetch of critical zones in the needed name hierarchy . . . . . . . . . . . . . . . . . . . . . . . . 6 3.3. .internal Local Zones . . . . . . . . . . . . . . . . . . 7 4. Zone Transfer Considerations . . . . . . . . . . . . . . . . 8 5. DNSSEC Considerations . . . . . . . . . . . . . . . . . . . . 8 6. ZoneMD Considerations . . . . . . . . . . . . . . . . . . . . 9 7. Network Operations Considerations . . . . . . . . . . . . . . 9 8. Deployment Scenarios . . . . . . . . . . . . . . . . . . . . 10 8.1. Crusing Spacecraft (CS) . . . . . . . . . . . . . . . . . 10 8.2. Moon Full Connectivity (LF) . . . . . . . . . . . . . . . 10 8.3. Moon Intermittent Connectivity (LI) . . . . . . . . . . . 10 9. Example Procedure . . . . . . . . . . . . . . . . . . . . . . 11 10. New TLD for Deep Space . . . . . . . . . . . . . . . . . . . 13 11. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 13 12. Security Considerations . . . . . . . . . . . . . . . . . . . 13 13. References . . . . . . . . . . . . . . . . . . . . . . . . . 13 13.1. Informative References . . . . . . . . . . . . . . . . . 13 Appendix A. Alternative Name Resolution Approaches . . . . . . . 15 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . 16 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 16 1. Introduction Deep space communications involve long delays (e.g., Earth to Mars has one-way delays 4-24 minutes) and intermittent communications, mainly because of orbital dynamics. To illustrate a simple example of intermittence, communications may be between a ground station on Earth to an asset on a planet through the use of an orbiter around that planet. That orbiter may have an overpass of 20 minutes every 2 hours. Therefore, continuous connectivity between the two endpoints cannot be assumed. However, as discussed in [I-D.many-tiptop-ip-architecture], orbiters may temporarily store IP packets until the next communication window with the next hop Blanchet & Hardaker Expires 19 May 2027 [Page 2] Internet-Draft DNS in Deep Space November 2026 appears. From the standpoint of the transport or upper application protocols, the one-way delay and the RTT will be stable while direct connectivity happens, and then jump to a large value while the orbiter is not reachable anymore, until it becomes reachable again. In some cases, where the asset is on a shadow side of a planet or Moon and while there is no full constellation of orbiters providing connectivity all around, the asset and Earth ground may never or rarely have a uninterrupted end2end path. [I-D.ietf-tiptop-usecase] provides more details about key characteristics of deep space communications and the use case of networking. [I-D.many-tiptop-ip-architecture] discusses the use of the whole IP stack in this context. Domain name requests and response over long delays generate timeouts and when there is no reachability to the DNS server, requests will not be answered. Therefore, on celestial bodies IP networks, a local DNS infrastructure with all the needed names and values stored locally is needed. While this document uses deep space as the base use case, it may apply to other "mostly" isolated networks, such as IoT networks. Mostly isolated means that most of the time the network is isolated, but there are times where it is not isolated and then may receive zone transfers or other means to populate or update its name caches. Various scenarios of deployments depending on the distance of the celestial body and the coverage of orbiters are described in [I-D.ietf-tiptop-usecase]. The first sections of this document describes on the hardest case (delays too long and intermittence) and then subsequently (Section 8) discusses more specific ones. 2. Use Case and Requirements Given all the investment in technology, protocol, services, infrastructure, scrutiny and governance in place, the ability to use the current Internet DNS Root with DNSSEC[RFC9364] in deep space has a lot of merit compared to using a split DNS tree with an alternate DNS root and trust chain. In order to achieve this goal, the necessary domain data required for full DNS resolution, including the necessary related DNSSEC keys, must be locally available for validation and use. The requirements of this document use case are: * given the inability to do real-time DNS queries to the Internet DNS infrastructure from space, ability to do real-time DNS queries using local infrastructure. Blanchet & Hardaker Expires 19 May 2027 [Page 3] Internet-Draft DNS in Deep Space November 2026 * domains used in deep space are under the Internet's unique DNS root[RFC2826], deployed under agreed community requirements[RFC7720] * ability to sometimes reach the celestial body network name servers from Earth Internet in order to update local caches of DNS data. * multiple network and DNS operators may exist on the celestial body network, each managing their own namespace and/or caches * management of DNS servers and zones needs to be possible from Earth, since no human is expected to do server administration on celestial body network * celestial body network autonomous DNS infrastructure that supports local resolution and DNSSEC validation 3. Possible Approaches This section presents various approaches that may meet the requirements set in the previous section. These approaches use a [RFC8806] (aka "LocalRoot" or "Hyperlocal Root") style approach for root zones, but augment it for the whole needed name hierarchy (many implementations of RFC8806 already support multiple zones). RFC8806 specifies an exact mechanism that a resolver should use when implementing a local copy of the root zone. However, existing resolver implementations do not support this exact mechanism, but instead implement the functionality using a number of internal methodologies. From a DNS client's perspective, the end result is that a recursive resolver supporting pre-fetching of the DNS Root or other zones fundamentally looks like the information was already available within the recursive resolver's cache. Some implementations actually do, in fact, simply fetch a zone file and then inject the results into its cache. For simplicity in the rest of this document, we'll refer to this concept as "pre-fetching a zone", regardless of how the implementation actually works internally. [Note: the DNSOP WG in the IETF is currently considering two potential update documents to RFC8806: draft-wkumari-dnsop-localroot- bcp and draft-hoffman-rootcache, though the call for adoption has not yet ended. This note will be removed by the authors (and hopefully obsolete) before publication.] Blanchet & Hardaker Expires 19 May 2027 [Page 4] Internet-Draft DNS in Deep Space November 2026 A important consideration is how long the data in the local implementation remains valid and cached. Implementations need to not expire entries from its cache that would degrade resolving queries in the target network. Additionally, celestial body resolver implementations should serve stale data when they are unable to obtain fresh data due to network failures [RFC8767]. "Target network" used in this document means the celestial body network. All approaches share similar naming infrastructure on the target celestial body network: * one or more recursive resolvers configured to mirror necessary authoritative zone data, such as the DNS root zone. * DNSSEC verifying resolvers will have the root trust anchor [trust-anchor] in their configuration, along with a cached copy of root zone. The authenticity root zone data will be verified using the ZONEMD record [RFC8976] * hosts need to be configured to use these recursive resolvers. It should be noted that, with proper planning and careful consideration, the following approaches are not exclusive and different techniques may be used in combination and deployed over time. 3.1. Pre-fetching of needed names and records If one assumes that all class, query names, and query types that will be used on the target network are known in advance, then recursively pre-fetching the contents of these needed records can be done on Internet and the responses saved in a file, together with the appropriate DNSSEC records, for loading into the cache of a recursive resolver. The records should contain all combinations of information that are relevant to the celestial body network. For example, an IP address record such as an A or AAAA record should resolve to an IP address relevant and reachable from the target network, or MX records for mail servers running within the celestial network. To achieve this, a list of all records fetched for needed query names, query types and query classes are saved and used as input for the above process. The resulting file containing all the records is uploaded to the caching recursive resolvers on the target network. Blanchet & Hardaker Expires 19 May 2027 [Page 5] Internet-Draft DNS in Deep Space November 2026 The recursive resolvers with pre-populated caches should require no further communication with any distant authoritative servers, or any of the domain tree records off planet, in order to resolve names within the target network. If a name used on the celestial body network by the hosts or applications is not in the uploaded file served by the local name servers, then the resolver handling the request will eventually timeout and return a SERVFAIL error. If all needed DNSSEC material is not fully uploaded, then DNSSEC validation will fail, also resulting in a SERVFAIL error. A method for syncing and updating all the updated records to the isolated network should be put in place, at the appropriate frequency. Typical DNS zone tranfer mechanisms (AXFR and IXFR) may not be suitable because TCP transport is likely not possible in deep space. In this case, other file transfer mechanisms may be used. Some DNS records have values containing other names, such as the SRV and CNAME records. The referenced names should also be included with the data to be fetched and cached by the pre-fetching recursive resolvers. Recursive resolvers that do not implement pre-fetching of required DNS data need to be configured as forwarding resolvers that forward requests to a recursive resolver that does. 3.2. Pre-fetch of critical zones in the needed name hierarchy If one assumes that the name hierarchy is known for all needed names used on the target network and if the operator of the DNS infrastructure on the target network has access to all the zones of the hierarchy, then the entire content of these zones can be pre- fetched and distributed to the target network's recursive resolvers. At a minimum, the DNS root zone should be included in this set of zones This approach have less risk of missing a name since all names under the hierarchy are uploaded. However, if the zones are too big compared to the transfer capacity to the target network, then this solution is not appropriate. Moreover, it may be possible that most of the names in the uploaded zones will not be used, therefore it is a possible waste of resources (bandwidth, memory/cpu on server, ...). Therefore, careful consideration of the chosen hierarchy, specially the top-level domains needed, is relevant. Given deep space relative limited use networks, it would make sense to dedicate some top-level domain or subdomain for mission needs (see the usage discussion of Blanchet & Hardaker Expires 19 May 2027 [Page 6] Internet-Draft DNS in Deep Space November 2026 .internal below). However, it is possible to remove all the non- needed record from the zones before uploading them to the target network DNS infrastructure, but then if some names are missing in this removal, the same issues from the previous approach appear. Note that DNSSEC validation will fail if the relevant validations records are removed from a signed zone. Importantly, the NSEC or NSEC3 in a signed zone are needed to support denial-of-existence for returning NXDOMAIN or NODATA answers to unanswerable queries. Recursive resolvers should make use of aggressive NSEC [RFC8198] for optimizing a local resolver's performance for query name and types which do not exist. A method for syncing and updating all the updated records to the target network should be put in place, at the appropriate frequency. Typical DNS zone tranfer mechanism may not be suitable because TCP transport is likely not possible in deep space. In this case, other file transfer mechanisms may be used. In the context of multiple operators on the target network, each one may do this process independently for its own zones, without having to rely on another party. 3.3. .internal Local Zones Authoritative local zones can be deployed on the target networks. However, in order to be secured by DNSSEC they will need a separate trust anchor for validation by resolvers making use of them. The danger of name collision with the Earth Internet DNS can create a big operational issues if not properly coordinated. ICANN's recommended mechanism for establishing locally controlled authoritative names is to make use of a sub-domain within the non- existent ".internal" TLD. Specifically, after a recommendation of the Security and Stability Advisory Committee ([sac113]), the ICANN board passed resolution 2024.07.29.06 ([internalresolution]) which set aside the .internal TLD for use within private networks, to ensure no conflicts would occur within the rest of the DNS tree. Conflicts may still arise within groups using the .internal TLD, and it is recommended that coordination occurs between missions such that each celestial network uses a sub-domain within the .internal non-TLD to further minimize collisions (for example "mars.internal"). Blanchet & Hardaker Expires 19 May 2027 [Page 7] Internet-Draft DNS in Deep Space November 2026 4. Zone Transfer Considerations If DNS zone transfer is possible over the link between the Internet and the target network, then incremental zone transfer (aka IXFR) might be advised to minimize the use of the bandwidth and also minimize the data merge on the target DNS server. Current zone transfer mechanisms, such as AXFR, IXFR, rsync or ftp, all use TCP as transport. However, TCP is not suitable for deep space. Therefore, it is recommended to use AXFR/IXFR over profiled[I-D.many-tiptop-quic-profile] QUIC[RFC9250]. If DNS zone transfer is not possible or not optimal, than another file transfer mechanism such as HTTP over profiled QUIC should be used. 5. DNSSEC Considerations Zones are signed at various frequencies based on the operator policies. If a signature on a record has expired, then DNSSEC validation will fail. The quantity that bounds how long the target network can operate while isolated is therefore the validity period of the RRSIG records covering the data it serves, that is the interval between the signature inception and expiration times, and not any TTL. The re-supply period from Earth should be shorter than the shortest signature validity period present in the hierarchy being served, with margin for missed contact opportunities. Similarly, the publication of any trust anchors in use and their intended replacement timelines, especially for the root zone anchor, should be monitored to make sure that new trust anchors are properly configured into resolvers before the old ones expire. The automated trust anchor update mechanism of [RFC5011] can only be relied on by validating resolvers that have pre-fetched the required information or recursive resolvers that are forwarding to pre- fetching recursive resolvers. Blanchet & Hardaker Expires 19 May 2027 [Page 8] Internet-Draft DNS in Deep Space November 2026 Those resolvers that are not configured to use local, up to date sources of zone data will not be able to functionally use RFC5011 in an isolated network. To support RFC5011, resolvers must be able to query the trust point at an interval derived from the DNSKEY TTL and the signature validity period. The current root zone parameters requires being able to send at least one query per day, to retry on failure at a comparable cadence, and to observe add and remove hold- down times of thirty days. A network that is out of contact for days or weeks cannot satisfy those timers. Trust anchors must therefore be established on Earth and delivered to the target network as part of the managed data set. Because trust anchors (typically just the DNS root) is the root of validation on the target network, it must not be accepted solely on the basis of the transfer that carried it. The receiving side should verify each key against the DS records published out of band by IANA [trust-anchor], by recomputing the digest of the DNSKEY as described in Section 5.1.4 of [RFC4034] and requiring it to match a published value before the key is installed as an anchor. This check uses only the published digests and the public key material itself, and so does not depend on the DNS that is about to be trusted. 6. ZoneMD Considerations Recursive resolvers that use pre-fetching of data to make use of locally cached DNS data, such as with the DNS root, must ensure that the ZONEMD record When glue records are changed in a zone, the ZoneMD [RFC8976] is checked within the zone data before using the contents of the retrieved zone. 7. Network Operations Considerations Even with careful management, there is some probability that some applications or hosts on the target network will query names that were pre-fetched by local resolvers on the target network, but refer to services or IP addresses that are not reachable from the target network. If the target network does have intermittent IP connectivity to Internet, but the link is not appropriate for live queries (because of long delays in deep space, expensive bandwidth, or very small time window of reachability, then the network may try to route DNS requests packets to the Internet which will likely result in timeouts and SERVFAIL responses. Therefore, a mechanisms to signal unreachability may be appropriate to be setup at the edge of the target network. Blanchet & Hardaker Expires 19 May 2027 [Page 9] Internet-Draft DNS in Deep Space November 2026 8. Deployment Scenarios Various scenarios of deployments depending on the distance of the celestial body and the coverage of orbiters are described in [I-D.ietf-tiptop-usecase]. This document describes the hardest cases. This section describes some specific ones that are typically easier. 8.1. Crusing Spacecraft (CS) This document assumes a minimal infrastructure on a celestial body surface and vicinty network. As discussed in [I-D.ietf-tiptop-usecase], there are other instances of IP networks in deep space. One of them is a cruising spacecraft which has non- continuous connectivity to Earth using a direct point-to-point link to Earth ground stations. A cruising spacecraft typically have a local IP network to connect the various payloads and computers together. Names may be used on the onboard network, either by a simple infrastructure as described in this document, or by other simpler means such as a static file distributed on all on-board computers. When the spacecraft lands or connects to an infrastructure network such as a celestial body surface network, some names may need to be updated to use the local DNS infrastructure. This transition is not discussed in this document. 8.2. Moon Full Connectivity (LF) A Moon surface network always reachable from Earth because of a sufficient number of orbiters enabling full connectivity, provides a simpler case, as DNS queries and responses could go through the deep space links with an RTT of less than 4 seconds. Resolvers will need to be configured with longer timeouts than traditional earth networks. DNS resolvers with caches on the surface network can then be used to optimize the name resolution traffic. However, if full connectivity is lost for some temporary windows, then DNS resolution may not work depending on the cache richness and RR TTLs. A more reliable infrastructure would be the one described in this document. 8.3. Moon Intermittent Connectivity (LI) A Moon surface network is reachable from Earth only during certain periods. Given an RTT of less than 4 seconds, DNS resolvers with caches on the surface network can then be used and DNS queries goes through the deep space links to Earth and responses back. If resource records TTL are lower than the longest intermittence interval, then those RR will be timeout in the cache unless serve stale [RFC8767] is configured for use, disabling name resolution for those names. Therefore, while this solution looks simpler, it is Blanchet & Hardaker Expires 19 May 2027 [Page 10] Internet-Draft DNS in Deep Space November 2026 fragile and not recommended. 9. Example Procedure This section gives an example procedure to build the minimal set of zones with which a resolver on the target network validates every name under a celestial zone against the Internet root trust anchor, without any query leaving the target network. It uses "moon.example.net" as the celestial zone, operated by the operator of "example.net", and M as the address of the authoritative name server on the target network. The zones to provision are the root zone, "net", "example.net" and "moon.example.net". Prerequisites: "example.net" is signed and its DS record is published in the "net" zone; the root trust anchor has been verified as described in Section 5 and is delivered to the target network. 1. The operator builds the zone "moon.example.net" on Earth, as a complete zone signed. It contains its SOA and NS records, the address M for its name server, and all the names used on the target network. Its signature validity period should exceed the longest expected isolation period, with margin. The operator computes its DS record. moon.example.net. SOA ns1.moon.example.net. hostmaster.example.net. (...) moon.example.net. NS ns1.moon.example.net. ns1.moon.example.net. A M rover1.moon.example.net. A
2. The operator builds the celestial edition of "example.net" on Earth and signs it with the same keys as the Earth edition, so that it validates under the DS record already published in "net". The address records of all the name servers listed at the apex are set to M. The edition contains the delegation of "moon.example.net" with its DS record and glue, and any other names of "example.net" used on the target network. The operator verifies that its key signing key still matches the DS record published in "net". example.net. SOA (...) example.net. NS ns1.example.net. ns1.example.net. A M example.net. DNSKEY