<?xml version="1.0" encoding="utf-8"?>
<!-- name="GENERATOR" content="github.com/mmarkdown/mmark Mmark Markdown Processor - mmark.miek.nl" -->
<rfc version="3" ipr="trust200902" docName="draft-ietf-dnsop-dns-catalog-zones-09" submissionType="IETF" category="std" xml:lang="en" xmlns:xi="http://www.w3.org/2001/XInclude" consensus="true">

<front>
<title abbrev="dns-catalog-zones">DNS Catalog Zones</title><seriesInfo value="draft-ietf-dnsop-dns-catalog-zones-09" stream="IETF" status="standard" name="Internet-Draft"></seriesInfo>
<author initials="P." surname="van Dijk" fullname="Peter van Dijk"><organization>PowerDNS</organization><address><postal><street></street>
<city>Den Haag</city>
<country>Netherlands</country>
</postal><email>peter.van.dijk@powerdns.com</email>
</address></author>
<author initials="L." surname="Peltan" fullname="Libor Peltan"><organization>CZ.NIC</organization><address><postal><street></street>
<country>CZ</country>
</postal><email>libor.peltan@nic.cz</email>
</address></author>
<author initials="O." surname="Sury" fullname="Ondrej Sury"><organization>Internet Systems Consortium</organization><address><postal><street></street>
<country>CZ</country>
</postal><email>ondrej@isc.org</email>
</address></author>
<author initials="W." surname="Toorop" fullname="Willem Toorop"><organization>NLnet Labs</organization><address><postal><street>Science Park 400</street>
<city>Amsterdam</city>
<code>1098 XH</code>
<country>Netherlands</country>
</postal><email>willem@nlnetlabs.nl</email>
</address></author>
<author initials="C.R." surname="Monshouwer" fullname="Kees Monshouwer"><organization></organization><address><postal><street></street>
<country>Netherlands</country>
</postal><email>mind@monshouwer.eu</email>
</address></author>
<author initials="P." surname="Thomassen" fullname="Peter Thomassen"><organization>deSEC, SSE - Secure Systems Engineering</organization><address><postal><street></street>
<city>Berlin</city>
<country>Germany</country>
</postal><email>peter@desec.io</email>
</address></author>
<author initials="A." surname="Sargsyan" fullname="Aram Sargsyan"><organization>Internet Systems Consortium</organization><address><postal><street></street>
</postal><email>aram@isc.org</email>
</address></author>
<date year="2023" month="February" day="7"></date>
<area>Internet</area>
<workgroup>DNSOP Working Group</workgroup>

<abstract>
<t>This document describes a method for automatic DNS zone provisioning among DNS
primary and secondary nameservers by storing and transferring the catalog of
zones to be provisioned as one or more regular DNS zones.</t>
</abstract>

</front>

<middle>

<section anchor="introduction"><name>Introduction</name>
<t>The content of a DNS zone is synchronized among its primary and secondary
nameservers using AXFR and IXFR.  However, the list of zones served by the
primary (called a catalog in <xref target="RFC1035"></xref>) is not automatically synchronized
with the secondaries.  To add or remove a zone, the administrator of a DNS
nameserver farm not only has to add or remove the zone from the primary, they
must also add/remove configuration for the zone from all secondaries.  This
can be both inconvenient and error-prone; in addition, the steps required are
dependent on the nameserver implementation.</t>
<t>This document describes a method in which the list of zones is represented as a
regular DNS zone (called a &quot;catalog zone&quot; here), and transferred using DNS zone
transfers.  When entries are added to or removed from the catalog zone, it is
distributed to the secondary nameservers just like any other zone.  Secondary
nameservers can then add/remove/modify the zones they serve in accordance with the
changes to the catalog zone. Other use-cases of nameserver remote configuration
by catalog zones are possible, where the catalog consumer might not be a
secondary.</t>
</section>

<section anchor="terminology"><name>Terminology</name>
<t>The key words &quot;<bcp14>MUST</bcp14>&quot;, &quot;<bcp14>MUST NOT</bcp14>&quot;, &quot;<bcp14>REQUIRED</bcp14>&quot;,
&quot;<bcp14>SHALL</bcp14>&quot;, &quot;<bcp14>SHALL NOT</bcp14>&quot;, &quot;<bcp14>SHOULD</bcp14>&quot;, &quot;<bcp14>SHOULD NOT</bcp14>&quot;,
&quot;<bcp14>RECOMMENDED</bcp14>&quot;, &quot;<bcp14>NOT RECOMMENDED</bcp14>&quot;, &quot;<bcp14>MAY</bcp14>&quot;, and
&quot;<bcp14>OPTIONAL</bcp14>&quot; in this document are to be interpreted as described in
BCP 14 <xref target="RFC2119"></xref><xref target="RFC8174"></xref> when, and only when, they appear in all
capitals, as shown here.</t>

<dl>
<dt>Catalog zone:</dt>
<dd>A DNS zone containing a DNS catalog, that is, a list of DNS zones and associated properties.</dd>
<dt>Member zone:</dt>
<dd>A DNS zone whose configuration is published inside a catalog zone.</dd>
<dt>Member node:</dt>
<dd>A DNS name in the Catalog zone representing a Member zone.</dd>
<dt><tt>$CATZ</tt>:</dt>
<dd>Used in examples as a placeholder to represent the domain name of the
catalog zone itself.
<tt>$OLDCATZ</tt> and <tt>$NEWCATZ</tt> are used to discuss migration of a member zone from one catalog zone <tt>$OLDCATZ</tt> to another catalog zone <tt>$NEWCATZ</tt>.</dd>
<dt>Catalog producer:</dt>
<dd>An entity that generates and is responsible for the contents of the catalog zone.</dd>
<dt>Catalog consumer:</dt>
<dd>An entity that extracts information from the catalog zone (such as a DNS
server that configures itself according to the catalog zone's contents).</dd>
</dl>
<t>This document makes use of terminology that is specific to the DNS, such as for transfer mechanisms (AXFR, IXFR), for record types (SOA, NS, PTR), and other technical terms (such as RDATA).
Since these terms have specific meanings in the DNS they are not expanded at first use in this document.
For definitions of those and other terms, see <xref target="RFC8499"></xref>.</t>
</section>

<section anchor="description"><name>Description</name>
<t>A catalog zone is a DNS zone whose contents are specially crafted. Its resource records (RR) primarily constitute a list of PTR records referencing other DNS zones (so-called &quot;member zones&quot;). The catalog zone may contain other records indicating additional metadata (so-called &quot;properties&quot;) associated with these member zones.</t>
<t>Catalog consumers MUST ignore any RRs in the catalog zone for which no processing is specified or which are otherwise not supported by the implementation.</t>
<t>Authoritative servers may be pre-configured with multiple catalog zones, each associated with a different set of configurations.</t>
<t>Although the contents of a catalog zone are interpreted and acted upon by
nameservers, a catalog zone is a regular DNS zone and so must adhere to the
standards for DNS zones.</t>
<t>A catalog zone is primarily intended for the management of a farm of authoritative nameservers, and should not be expected to be accessible from any recursive nameserver.</t>
</section>

<section anchor="catalog-zone-structure"><name>Catalog Zone Structure</name>
<t>A catalog zone MUST follow the usual rules for DNS zones.
In particular, SOA and NS record sets MUST be present and adhere to standard requirements (such as <xref target="RFC1982"></xref>).</t>
<t>Although catalog zones are not intended to be queried via recursive resolution (see <xref target="security"></xref>), at least one NS RR is still required so that a catalog zone is a syntactically correct DNS zone.
A single NS RR with a NSDNAME field containing the absolute name &quot;invalid.&quot; is RECOMMENDED <xref target="RFC2606"></xref><xref target="RFC6761"></xref>.</t>

<section anchor="listofmemberzones"><name>Member Zones</name>
<t>The list of member zones is specified as a collection of member nodes, represented by domain names under the owner name &quot;zones&quot; where &quot;zones&quot; is a direct child domain of the catalog zone.</t>
<t>The names of member zones are represented on the RDATA side (instead of as a part of owner names) of a PTR record, so that all valid domain names may be represented regardless of their length <xref target="RFC1035"></xref>.
This PTR record MUST be the only record in the PTR RRset with the same name.
The presence of more than one record in the RRset indicates a broken catalog zone which MUST NOT be processed (see <xref target="generalrequirements"></xref>).</t>
<t>For example, if a catalog zone lists three zones &quot;example.com.&quot;,
&quot;example.net.&quot; and &quot;example.org.&quot;, the member node RRs would appear as follows:</t>

<artwork>&lt;unique-1&gt;.zones.$CATZ 0 IN PTR example.com.
&lt;unique-2&gt;.zones.$CATZ 0 IN PTR example.net.
&lt;unique-3&gt;.zones.$CATZ 0 IN PTR example.org.
</artwork>
<t>where <tt>&lt;unique-N&gt;</tt> is a label that tags each record in the collection.
<tt>&lt;unique-N&gt;</tt> has a unique value in the collection.
When different <tt>&lt;unique-N&gt;</tt> labels hold the same PTR value (i.e., point to the same member zone), the catalog zone is broken and MUST NOT be processed (see <xref target="generalrequirements"></xref>).</t>
<t>Member node labels carry no informational meaning beyond labeling member zones.
A changed label may indicate that the state for a zone needs to be reset (see <xref target="zonereset"></xref>).</t>
<t>Having the zones uniquely tagged with the <tt>&lt;unique-N&gt;</tt> label ensures that additional RRs can be added below the member node (see <xref target="properties"></xref>).</t>
<t>The CLASS field of every RR in a catalog zone MUST be IN (1).
The TTL field's value has no meaning in this context and SHOULD be ignored.</t>
</section>

<section anchor="properties"><name>Properties</name>
<t>Catalog zone information is stored in the form of &quot;properties&quot;.</t>
<t>Properties are identified by their name, which is used as an owner name prefix for one or more record sets underneath a member node (or underneath the catalog zone apex), with RR type(s) as appropriate for the respective property.</t>
<t>Known properties with the correct RR type, but which are for some reason
invalid (for example because of an impossible value or because of an illegal
number of RRs in the RRset), denote a broken catalog zone which MUST NOT be
processed (see <xref target="generalrequirements"></xref>).</t>
<t>This document includes a set of initial properties which can be extended via the IANA registry defined and created in <xref target="iana"></xref>.
Some properties are defined at the global level; others are scoped to apply only to a specific member zone.
This document defines a mandatory global property in <xref target="version"></xref>.
The &quot;zones&quot; label from <xref target="listofmemberzones"></xref> can also be seen as a global property and is listed as such in the IANA registry in <xref target="iana"></xref>.
Member-specific properties are described in <xref target="memberproperties"></xref>.</t>
<t>Implementers may store additional information in the catalog zone with Custom properties, see <xref target="customproperties"></xref>.
The meaning of such custom properties is determined by the implementation in question.</t>

<section anchor="version"><name>Schema Version (<tt>version</tt> property)</name>
<t>The catalog zone schema version is specified by an integer value embedded in a TXT RR named <tt>version.$CATZ</tt>.
All catalog zones MUST have a TXT RRset named <tt>version.$CATZ</tt> with exactly one RR.</t>
<t>Catalog consumers MUST NOT apply catalog zone processing to</t>

<ul>
<li>zones without the <tt>version</tt> property</li>
<li>zones with a <tt>version</tt> property with more than one RR in the RRset</li>
<li>zones with a <tt>version</tt> property without an expected value in the
<tt>version.$CATZ</tt> TXT RR</li>
<li>zones with a <tt>version</tt> property with a schema version value which is not implemented by the consumer (e.g. version &quot;1&quot;)</li>
</ul>
<t>These conditions signify a broken catalog zone which MUST NOT be processed (see
<xref target="generalrequirements"></xref>).</t>
<t>For this memo, the value of the <tt>version.$CATZ</tt> TXT RR MUST be set to &quot;2&quot;, i.e.:</t>

<sourcecode type="dns-zone">version.$CATZ 0 IN TXT &quot;2&quot;
</sourcecode>
<t>NB: Version 1 was used in a draft version of this memo and reflected
the implementation first found in BIND 9.11.</t>
</section>
</section>

<section anchor="memberproperties"><name>Member Zone Properties</name>
<t>Each member zone MAY have one or more additional properties, described in this chapter.
The member properties described in this document are all optional and implementations MAY choose to implement all, some or none of them.
Member zone properties are represented by RRsets below the corresponding member node.</t>

<section anchor="cooproperty"><name>Change of Ownership (<tt>coo</tt> property)</name>
<t>The <tt>coo</tt> property facilitates controlled migration of a member zone from one catalog to another.</t>
<t>A Change Of Ownership is signaled by the <tt>coo</tt> property in the catalog zone currently &quot;owning&quot; the zone.
The name of the new catalog is the value of a PTR record in the relevant coo property in the old catalog.
For example if member &quot;example.com.&quot; will migrate from catalog zone <tt>$OLDCATZ</tt> to catalog zone <tt>$NEWCATZ</tt>, this appears in the <tt>$OLDCATZ</tt> catalog zone as follows:</t>

<artwork>&lt;unique-N&gt;.zones.$OLDCATZ 0 IN PTR example.com.
coo.&lt;unique-N&gt;.zones.$OLDCATZ 0 IN PTR $NEWCATZ
</artwork>
<t>The PTR RRset MUST consist of a single PTR record.
The presence of more than one record in the RRset indicates a broken catalog zone which MUST NOT be processed (see <xref target="generalrequirements"></xref>).</t>
<t>When a consumer of a catalog zone <tt>$OLDCATZ</tt> receives an update which adds or changes a <tt>coo</tt> property for a member zone in <tt>$OLDCATZ</tt>, it does <em>not</em> migrate the member zone immediately.
The migration has to wait for an update of <tt>$NEWCATZ</tt>. in which the member zone is present. The consumer MUST verify, before the actual migration, that <tt>coo</tt> property pointing to <tt>$NEWCATZ</tt> is still present in <tt>$OLDCATZ</tt>.</t>
<t>Unless the member node label (i.e., <tt>&lt;unique-N&gt;</tt>) for the member is the same in <tt>$NEWCATZ</tt>, all its associated state for a just migrated zone MUST be reset (see <xref target="zonereset"></xref>).
Note that the owner of <tt>$OLDCATZ</tt> allows for the zone associated state to be taken over by the owner of <tt>$NEWCATZ</tt> by default.
To prevent the takeover of state, the owner of <tt>$OLDCATZ</tt> must remove this state by updating the associated properties or by performing a zone state reset (see <xref target="zonereset"></xref>) before or simultaneous with adding the <tt>coo</tt> property. (see also <xref target="security"></xref>)</t>
<t>The old owner may remove the member zone containing the <tt>coo</tt> property from <tt>$OLDCATZ</tt> once it has been established that all its consumers have processed the Change of Ownership.</t>
</section>

<section anchor="groups-group-property"><name>Groups (<tt>group</tt> property)</name>
<t>With a <tt>group</tt> property, consumer(s) can be signaled to treat some member zones within the catalog zone differently.</t>
<t>The consumer MAY apply different configuration options when processing member zones, based on the value of the <tt>group</tt> property.
A <tt>group</tt> property value is stored as the entire RDATA of a TXT record directly below the member node.
The exact handling of the <tt>group</tt> property value is left to the consumer's implementation and configuration.</t>
<t>The producer MAY assign a <tt>group</tt> property to all, some, or none of the member zones within a catalog zone.
The producer MAY assign more than one <tt>group</tt> property to one member zone. This will make it possible to transfer group information for different consumer operators in a single catalog zone.
Implementations MAY facilitate mapping of a specific <tt>group</tt> value to specific configuration configurable <em>on a per catalog zone basis</em> to allow for producers that publish their catalog zone at multiple consumer operators.
Consumer operators SHOULD namespace their group values to reduce the risk of having to resolve clashes.</t>
<t>The consumer MUST ignore <tt>group</tt> values it does not understand.
When a consumer encounters multiple <tt>group</tt> values for a single member zone, it MAY choose to process all, some or none of them. This is left to the implementation.</t>

<section anchor="example"><name>Example</name>
<t>Group properties are represented by TXT resource records. The record content
has no pre-defined meaning. Their interpretation is purely a matter
of agreement between the producer and the consumer(s) of the catalog.</t>
<t>For example, the &quot;foo&quot; group could be agreed to indicate that a zone not
be signed with DNSSEC. Conversely, an agreement could define that group names
starting with &quot;operator-&quot; indicate in which way a given DNS operator should set
up certain aspects of the member zone's DNSSEC configuration.</t>
<t>Assuming that the catalog producer and consumer(s) have established such
agreements, consider the following catalog zone (snippet) which signals to
consumer(s) how to treat DNSSEC for the zones &quot;example.net.&quot; and &quot;example.com.&quot;:</t>

<artwork>&lt;unique-1&gt;.zones.$CATZ        0 IN PTR    example.com.
group.&lt;unique-1&gt;.zones.$CATZ  0 IN TXT    &quot;foo&quot;
&lt;unique-2&gt;.zones.$CATZ        0 IN PTR    example.net.
group.&lt;unique-2&gt;.zones.$CATZ  0 IN TXT    &quot;operator-x-foo&quot;
group.&lt;unique-2&gt;.zones.$CATZ  0 IN TXT    &quot;operator-y&quot; &quot;bar&quot;

</artwork>
<t>In this scenario, consumer(s) shall, by agreement, not sign the member zone &quot;example.com.&quot; with DNSSEC.
For &quot;example.net.&quot;, the consumers, at two different operators, will configure
the member zone to be signed with a specific combination of settings. The group value that indicates
that depends on what has been agreed with each operator (&quot;operator-x-foo&quot; vs. &quot;operator-y&quot; &quot;bar&quot;).</t>
</section>
</section>
</section>

<section anchor="customproperties"><name>Custom Properties (<tt>*.ext</tt> properties)</name>
<t>Implementations and operators of catalog zones may choose to provide their own properties.
Custom properties can occur both globally, or for a specific member zone.
To prevent a name clash with future properties, such properties MUST be represented below the label &quot;ext&quot;.</t>
<t>&quot;ext&quot; is not a placeholder. A custom property is named as follows:</t>

<artwork>; a global custom property:
&lt;property-prefix&gt;.ext.$CATZ

; a member zone custom property:
&lt;property-prefix&gt;.ext.&lt;unique-N&gt;.zones.$CATZ
</artwork>
<t><tt>&lt;property-prefix&gt;</tt> may consist of one or more labels.</t>
<t>Implementations SHOULD namespace their custom properties to limit risk of clashes with other implementations of catalog zones.
This can be achieved by using two labels as the <tt>&lt;property-prefix&gt;</tt>, so that the
name of the implementation is included in the prefix: <tt>&lt;some-setting&gt;.&lt;implementation-name&gt;.ext.$CATZ</tt>.</t>
<t>Implementations MAY use such properties on the member zone level to store additional information about member zones,
for example to flag them for specific treatment.</t>
<t>Further, implementations MAY use custom properties on the global level to store additional information about the catalog zone itself.
While there may be many use cases for this, a plausible one is to store default values for custom properties on the global level,
then overriding them using a property of the same name on the member level (= under the <tt>ext</tt> label of the member node) if so desired.
A property agreement between producer and consumer should clearly define what semantics apply, and whether a property is global, member, or both.</t>
<t>The meaning of the custom properties described in this section is determined by the implementation alone, without expectation of interoperability.</t>
</section>
</section>

<section anchor="behavior"><name>Nameserver Behavior</name>

<section anchor="generalrequirements"><name>General Requirements</name>
<t>As it is a regular DNS zone, a catalog zone can be transferred using DNS zone
transfers among nameservers.</t>
<t>Catalog updates should be automatic, i.e., when a nameserver that supports
catalog zones completes a zone transfer for a catalog zone, it SHOULD apply
changes to the catalog within the running nameserver automatically without any
manual intervention.</t>
<t>Nameservers MAY allow loading and transfer of broken zones with incorrect
catalog zone syntax (as they are treated as regular zones).
The reason a catalog zone is considered broken SHOULD be communicated clearly to the operator (e.g. through a log message).</t>
<t>When a previously correct catalog zone becomes a broken catalog zone, because
of an update through an incremental transfer or otherwise, it loses its catalog
meaning.
No special processing occurs. Member zones previously configured by this catalog
MUST NOT be removed or reconfigured in any way.</t>
<t>If a name server restarts with a broken catalog zone, the broken catalog SHOULD
NOT prevent the name server from starting up and serving the member zones in
the last valid version of the catalog zone.</t>
<t>Processing of a broken catalog SHALL start (or resume) when the catalog turns
into a correct catalog zone, for example by an additional update (through zone
transfer or updates) fixing the catalog zone.</t>
<t>Similarly, when a catalog zone expires, it loses its catalog meaning and
MUST no longer be processed as such.
No special processing occurs until the zone becomes fresh again.</t>
</section>

<section anchor="nameclash"><name>Member zone name clash</name>
<t>If there is a clash between an existing zone's name (either from an existing member zone or otherwise configured zone) and an incoming
member zone's name (via transfer or update), the new instance of the zone MUST
be ignored and an error SHOULD be logged.</t>
<t>A clash between an existing member zone's name and an incoming member zone's name (via transfer or update), may be an attempt to migrate a zone to a different catalog, but should not be treated as one except as described in <xref target="cooproperty"></xref>.</t>
</section>

<section anchor="zoneremoval"><name>Member zone removal</name>
<t>When a member zone is removed from a specific catalog zone, a consumer MUST NOT remove the zone and associated state data if the zone was not configured from that specific catalog zone.
Only when the zone was configured from a specific catalog zone, and the zone is removed as a member from that specific catalog zone, the zone and associated state (such as zone data and DNSSEC keys) MUST be removed from the consumer.
Consumer operators may consider to temporarily archive associated state to facilitate mistake recovery.</t>
</section>

<section anchor="namechange"><name>Member node name change</name>
<t>When via a single update or transfer, the member node's label value (<tt>&lt;unique-N&gt;</tt>) changes, catalog consumers MUST process this as a member zone removal including all the zone's associated state (as described in <xref target="zoneremoval"></xref>), immediately followed by processing the member as a newly to be configured zone in the same catalog.
</t>
</section>

<section anchor="zonemigration"><name>Migrating member zones between catalogs</name>
<t>If all consumers of the catalog zones involved support the <tt>coo</tt> property, it is RECOMMENDED to perform migration of a member zone by following the procedure described in <xref target="cooproperty"></xref>.
Otherwise, a migration of a member zone from a catalog zone <tt>$OLDCATZ</tt> to a catalog zone <tt>$NEWCATZ</tt> has to be done by: first removing the member zone from <tt>$OLDCATZ</tt>; second adding the member zone to <tt>$NEWCATZ</tt>.</t>
<t>If in the process of a migration some consumers of the involved catalog zones did not catch the removal of the member zone from <tt>$OLDCATZ</tt> yet (because of a lost packet or downtime or otherwise), but did already see the update of <tt>$NEWCATZ</tt>, they may consider the update adding the member zone in <tt>$NEWCATZ</tt> to be a name clash (see <xref target="nameclash"></xref>) and as a consequence the member is not migrated to <tt>$NEWCATZ</tt>.
This possibility needs to be anticipated with a member zone migration.
Recovery from such a situation is out of the scope of this document.
It may for example entail a manually forced retransfer of <tt>$NEWCATZ</tt> to consumers after they have been detected to have received and processed the removal of the member zone from <tt>$OLDCATZ</tt>.</t>
</section>

<section anchor="zonereset"><name>Zone-associated state reset</name>
<t>It may be desirable to reset state (such as zone data and DNSSEC keys) associated with a member zone.</t>
<t>A zone state reset may be performed by a change of the member node's name (see <xref target="namechange"></xref>).</t>
</section>
</section>

<section anchor="implementationnotes"><name>Implementation and Operational Notes</name>
<t>Although any valid domain name can be used for the catalog name $CATZ, a catalog producer MUST NOT use names that are not under the control of the catalog producer (with the exception of reserved names). It is
RECOMMENDED to use either a domain name owned by the catalog producer, or
to use a name under a suitable name such as &quot;invalid.&quot; <xref target="RFC6761"></xref>.</t>
<t>Catalog zones on secondary nameservers would have to be set up manually, perhaps
as static configuration, similar to how ordinary DNS zones are configured when catalog zones or another automatic configuration mechanism are not in place.
The secondary additionally needs to be configured as a catalog consumer for the catalog zone to enable processing of the member zones in the catalog, such as automatic synchronization of the member zones for secondary service.</t>
<t>Operators of catalog consumers should note that secondary name servers may
receive DNS NOTIFY messages <xref target="RFC1996"></xref> for zones before they are seen as
newly added member zones to the catalog from which that secondary is
provisioned.</t>
<t>Although they are regular DNS zones, catalog zones contain only information for
the management of a set of authoritative nameservers.  To prevent unintended
exposure to other parties, operators SHOULD limit the systems able to query these zones.</t>
<t>Querying/serving catalog zone contents may be inconvenient via DNS
due to the nature of their representation.
An administrator may therefore want to use a different method for
looking at data inside the catalog zone.  Typical
queries might include dumping the list of member zones, dumping a member zone's
effective configuration, querying a specific property value of a member zone,
etc.  Because of the structure of catalog zones, it may not be possible to
perform these queries intuitively, or in some cases, at all, using DNS QUERY.
For example, it is not possible to enumerate the contents of a multivalued
property (such as the list of member zones) with a single QUERY.
Implementations are therefore advised to provide a tool that uses either the
output of AXFR or an out-of-band method to perform queries on catalog zones.</t>
<t>Great power comes with great responsibility: Catalog zones simplify zone
provisioning by orchestrating zones on secondary name servers from a single
data source - the catalog.  Hence, the catalog producer has great power and
changes must be treated carefully. For example if the catalog is generated by
some script and this script for whatever reason generates an empty catalog,
millions of member zones may get deleted from their secondaries within seconds
and all the affected domains may be offline in a blink of an eye.</t>
</section>

<section anchor="security"><name>Security Considerations</name>
<t>As catalog zones are transmitted using DNS zone transfers,
it is RECOMMENDED that catalog zone transfers are protected from unexpected modifications by way of authentication,
for example by using TSIG <xref target="RFC8945"></xref>, or Strict or Mutual TLS authentication with DNS Zone transfer over TLS or QUIC <xref target="RFC9103"></xref>.</t>
<t>Use of DNS UPDATE <xref target="RFC2136"></xref> to modify the content of catalog zones SHOULD similarly be authenticated.</t>
<t>Zone transfers of member zones SHOULD similarly be authenticated.
TSIG shared secrets used for member zones SHOULD NOT be mentioned in the catalog zone data.
However, key identifiers may be shared within catalog zones.</t>
<t>Catalog zones reveal the zones served by their consumers, including their properties.
To prevent unintentional exposure of catalog zone contents, it is RECOMMENDED to limit the systems able to query them and to conduct catalog zone transfers confidentially <xref target="RFC9103"></xref>.</t>
<t>As with regular zones, primary and secondary nameservers for a catalog zone may
be operated by different administrators.  The secondary nameservers may be
configured as a catalog consumer to synchronize catalog zones from the primary, but the primary's
administrators may not have any administrative access to the secondaries.</t>
<t>Administrative control over what zones are served from the configured name servers shifts completely from the server operator (consumer) to the &quot;owner&quot; (producer) of the catalog zone content.
To prevent unintended provisioning of zones, consumer(s) SHOULD scope the set of
admissible member zones by any means deemed suitable (such as statically, via
regular expressions, or dynamically, by verifying against another database
before accepting a member zone).</t>
<t>With migration of member zones between catalogs using the <tt>coo</tt> property, it is possible for the owner of the target catalog (i.e., <tt>$NEWCATZ</tt>) to take over all its associated state with the zone from the original owner (i.e., <tt>$OLDCATZ</tt>) by maintaining the same member node label (i.e., <tt>&lt;unique-N&gt;</tt>).
To prevent the takeover of the zone associated state, the original owner has to enforce a zone state reset by changing the member node label (see <xref target="zonereset"></xref>) before or simultaneously with adding the <tt>coo</tt> property.</t>
</section>

<section anchor="iana"><name>IANA Considerations</name>
<t>IANA is requested to create a registry on the &quot;Domain Name System (DNS) Parameters&quot; IANA web page as follows:</t>

<dl>
<dt>Registry Name:</dt>
<dd>DNS Catalog Zones Properties</dd>
<dt>Assignment Policy:</dt>
<dd>Expert Review, except for property prefixes ending in the label &quot;ext&quot;, which are for Private Use.</dd>
<dt>Reference:</dt>
<dd>[this document]</dd>
<dt>Note:</dt>
<dd>This registry does not apply to Catalog Zones version &quot;1&quot;, but applies to Catalog Zones version &quot;2&quot; as specified in [this document].</dd>
</dl>
<table>
<thead>
<tr>
<th>Property prefix</th>
<th>Description</th>
<th>Status</th>
<th>Reference</th>
</tr>
</thead>

<tbody>
<tr>
<td>zones</td>
<td>List of member zones</td>
<td>Standards Track</td>
<td>[this document]</td>
</tr>

<tr>
<td>version</td>
<td>Schema version</td>
<td>Standards Track</td>
<td>[this document]</td>
</tr>

<tr>
<td>coo</td>
<td>Change of Ownership</td>
<td>Standards Track</td>
<td>[this document]</td>
</tr>

<tr>
<td>group</td>
<td>Group</td>
<td>Standards Track</td>
<td>[this document]</td>
</tr>

<tr>
<td>*.ext</td>
<td>Custom properties</td>
<td>Private Use</td>
<td>[this document]</td>
</tr>
</tbody>
</table><t>The meanings of the fields are as follows:</t>

<dl>
<dt>Property prefix:</dt>
<dd>One or more domain name labels</dd>
<dt>Description:</dt>
<dd>A human readable short description or name for the property</dd>
<dt>Status:</dt>
<dd>IETF Document status or &quot;External&quot; if not documented in an IETF document.</dd>
<dt>Reference:</dt>
<dd>A stable reference to the document in which this property is defined.</dd>
</dl>
</section>

<section anchor="acknowledgements"><name>Acknowledgements</name>
<t>Our deepest thanks and appreciation go to Stephen Morris,
Ray Bellis and Witold Krecicki who initiated this draft and did the bulk of
the work.</t>
<t>Catalog zones originated as the chosen method among various proposals that were
evaluated at ISC for easy zone management.  The chosen method of storing the
catalog as a regular DNS zone was proposed by Stephen Morris.</t>
<t>The initial authors discovered that Paul Vixie's earlier <xref target="Metazones"></xref> proposal
implemented a similar approach and reviewed it.  Catalog zones borrow some
syntax ideas from Metazones, as both share this scheme of representing the
catalog as a regular DNS zone.</t>
<t>Thanks to Leo Vandewoestijne. Leo's presentation in the DNS devroom at the
FOSDEM'20 <xref target="FOSDEM20"></xref> was one of the motivations to take up and continue the
effort of standardizing catalog zones.</t>
<t>Thanks to Joe Abley, David Blacka, Brian Conry, Klaus Darilion, Brian Dickson, Tony Finch, Evan Hunt, Shane Kerr, Warren Kumari, Patrik Lundin, Matthijs Mekking, Victoria Risk, Josh Soref, Petr Spacek, Michael StJohns, Carsten Strotmann and Tim Wicinski for reviewing draft proposals and offering comments and suggestions.</t>
</section>

</middle>

<back>
<references><name>Normative References</name>
<xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.1035.xml"/>
<xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.1982.xml"/>
<xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.1996.xml"/>
<xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2119.xml"/>
<xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2136.xml"/>
<xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.2606.xml"/>
<xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.6761.xml"/>
<xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.8174.xml"/>
<xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.8499.xml"/>
<xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.8945.xml"/>
<xi:include href="https://xml2rfc.ietf.org/public/rfc/bibxml/reference.RFC.9103.xml"/>
</references>
<references><name>Informative References</name>
<reference anchor="FOSDEM20" target="https://archive.fosdem.org/2020/schedule/event/dns_catz/">
  <front>
    <title>Extending Catalog zones - another approach in automating maintenance</title>
    <author fullname="Leo Vandewoestijne" initials="L." surname="Vandewoestijne"></author>
    <date year="2020"></date>
  </front>
</reference>
<reference anchor="Metazones" target="http://family.redbarn.org/~vixie/mz.pdf">
  <front>
    <title>Federated Domain Name Service Using DNS Metazones</title>
    <author fullname="Paul Vixie" initials="P." surname="Vixie"></author>
    <date year="2005"></date>
  </front>
</reference>
</references>

<section anchor="catalog-zone-example"><name>Catalog Zone Example</name>
<t>The following is a full example of a catalog zone containing three member zones with various properties:</t>

<artwork>catalog.invalid.                                0  SOA   invalid. (
                        invalid. 1625079950 3600 600 2147483646 0 )
catalog.invalid.                                0  NS    invalid.
example.vendor.ext.catalog.invalid.             0  CNAME example.net.
version.catalog.invalid.                        0  TXT   &quot;2&quot;
nj2xg5b.zones.catalog.invalid.                  0  PTR   example.com.
nvxxezj.zones.catalog.invalid.                  0  PTR   example.net.
group.nvxxezj.zones.catalog.invalid.            0  TXT   (
                        &quot;operator-x-foo&quot; )
nfwxa33.zones.catalog.invalid.                  0  PTR   example.org.
coo.nfwxa33.zones.catalog.invalid.              0  PTR   (
                        newcatz.invalid. )
group.nfwxa33.zones.catalog.invalid.            0  TXT   (
                        &quot;operator-y-bar&quot; )
metrics.vendor.ext.nfwxa33.zones.catalog.invalid. 0  CNAME (
                        collector.example.net. )
</artwork>
</section>

<section anchor="implementation-status"><name>Implementation Status</name>
<t><strong>Note to the RFC Editor</strong>: please remove this entire appendix before publication.</t>
<t>In the following implementation status descriptions, &quot;DNS Catalog Zones&quot; refers
to DNS Catalog Zones version 2 as described in this document.
Version 1 of catalog zones was initially developed by ISC for BIND, but never standardized in the IETF.
Support for version 1 catalog zones is explicitly mentioned per implementation.
Support for the <tt>coo</tt> and <tt>group</tt> properties are also explicitly mentioned per implementation.</t>

<ul>
<li><t>Knot DNS 3.1 (released August 2, 2021) supports both producing and consuming of catalog zones, including the group property.</t>
</li>
<li><t>PowerDNS from version 4.7 (released October 3, 2022) supports both producing and consuming of catalog zones version 2 and consuming of catalog zones version 1.
PowerDNS does support the <tt>coo</tt> property, and the <tt>group</tt> property on the producing side.</t>
</li>
<li><t>Proof of concept <eref target="https://github.com/IETF-Hackathon/NSDCatZ">python scripts</eref>
that can be used for both generating and consuming DNS Catalog Zones with NSD
have been developed during the hackathon at the IETF-109.</t>
</li>
<li><t>BIND 9.18.3+ supports version 2 catalog zones as described in this document including the <tt>coo</tt> property, as well as version 1 catalog zones.</t>
</li>
</ul>
<t>Interoperability between the above implementations has been tested during the
hackathon at the IETF-109.</t>
</section>

<section anchor="change-history"><name>Change History</name>
<t><strong>Note to the RFC Editor</strong>: please remove this entire appendix before publication.</t>

<ul>
<li>draft-muks-dnsop-dns-catalog-zones-00</li>
</ul>
<blockquote><t>Initial public draft.</t>
</blockquote>
<ul>
<li>draft-muks-dnsop-dns-catalog-zones-01</li>
</ul>
<blockquote><t>Added Witold, Ray as authors.  Fixed typos, consistency issues.
  Fixed references.  Updated Area.  Removed newly introduced custom
  RR TYPEs.  Changed schema version to 1.  Changed TSIG requirement
  from MUST to SHOULD.  Removed restrictive language about use of
  DNS QUERY.  When zones are introduced into a catalog zone, a
  primary SHOULD first make the new zones available for transfers
  first (instead of MUST).  Updated examples, esp. use IPv6 in
  examples per Fred Baker.  Add catalog zone example.</t>
</blockquote>
<ul>
<li>draft-muks-dnsop-dns-catalog-zones-02</li>
</ul>
<blockquote><t>Addressed some review comments by Patrik Lundin.</t>
</blockquote>
<ul>
<li>draft-muks-dnsop-dns-catalog-zones-03</li>
</ul>
<blockquote><t>Revision bump.</t>
</blockquote>
<ul>
<li>draft-muks-dnsop-dns-catalog-zones-04</li>
</ul>
<blockquote><t>Reordering of sections into more logical order.
  Separation of multi-valued properties into their own category.</t>
</blockquote>
<ul>
<li>draft-toorop-dnsop-dns-catalog-zones-00</li>
</ul>
<blockquote><t>New authors to pickup the editor pen on this draft</t>
<t>Remove data type definitions for zone properties
  Removing configuration of member zones through zone properties altogether</t>
<t>Remove Open issues and discussion Appendix, which was about zone options (including primary/secondary relationships) only.</t>
</blockquote>
<ul>
<li>draft-toorop-dnsop-dns-catalog-zones-01</li>
</ul>
<blockquote><t>Added a new section &quot;The Serial Property&quot;, introducing a new mechanism
  which can help with disseminating zones from the primary to the secondary
  nameservers in a timely fashion more reliably.</t>
<t>Three different ways to provide a &quot;serial&quot; property with a member zone are
  offered to or the workgroup for discussion.</t>
<t>Added a new section &quot;Implementation Status&quot;, listing production ready,
  upcoming and Proof of Concept implementations, and reporting on
  interoperability of the different implementations.</t>
</blockquote>
<ul>
<li>draft-toorop-dnsop-dns-catalog-zones-02</li>
</ul>
<blockquote><t>Adding the <tt>coo</tt> property for zone migration in a controlled fashion</t>
<t>Adding the <tt>group</tt> property for reconfigure settings of member zones
  in an atomic update</t>
<t>Adding the <tt>epoch</tt> property to reset zone associated state in a controlled fashion</t>
</blockquote>
<ul>
<li>draft-toorop-dnsop-dns-catalog-zones-03</li>
</ul>
<blockquote><t>Big cleanup!</t>
<t>Introducing the terms catalog consumer and catalog producer</t>
<t>Reorganized topics to create a more coherent whole</t>
<t>Properties all have consistent format now</t>
<t>Try to assume the least possible from implementations w.r.t.:</t>
<t>1) Predictability of the &lt;unique-N&gt; IDs of member zones</t>
<t>2) Whether or not fallback catalog zones can be found for a member</t>
<t>3) Whether or not a catalog consumer can maintain state</t>
</blockquote>
<ul>
<li>draft-toorop-dnsop-dns-catalog-zones-04</li>
</ul>
<blockquote><t>Move Implementation status to appendix</t>
<t>Miscellaneous textual improvements</t>
<t><tt>coo</tt> property points to <tt>$NEWCATZ</tt> (and not <tt>zones.$NEWCATZ</tt>)</t>
<t>Remove suggestion to increase serial and remove member zone from <tt>$OLDCATZ</tt> after migration</t>
<t>More consistent usage of the terms catalog consumer and catalog producer throughout the document</t>
<t>Better (safer) description of resetting refresh timers of member zones with the <tt>serial</tt> property</t>
<t>Removing a member MUST remove zone associated state</t>
<t>Make authentication requirements a bit less prescriptive in security considerations</t>
<t>Updated implementation status for KnotDNS</t>
<t>Describe member node name changes and update &quot;Zone associated state reset&quot; to use that as the mechanism for it.</t>
<t>Add Peter Thomassen as co-author</t>
<t>Complete removal of the <tt>epoch</tt> property.  We consider consumer optimizations with predictable member node labels (for example based on a hash) out of the scope of this document.</t>
<t>Miscellaneous editorial improvements</t>
</blockquote>
<ul>
<li>draft-toorop-dnsop-dns-catalog-zones-05</li>
</ul>
<blockquote><t>Add Kees Monshouwer as co-author</t>
<t>Removed the &quot;serial&quot; property</t>
<t>Allow custom properties on the global level</t>
</blockquote>
<ul>
<li>draft-toorop-dnsop-dns-catalog-zones-06</li>
</ul>
<blockquote><t>Move administrative control explanation to Security Considerations</t>
<t>Move comment on query methods to Implementation Notes</t>
<t>Clarify what happens on expiry</t>
<t>Clarify catalog consumer behavior when MUST condition is violated</t>
<t>Better text on ordering of operations for Change of Ownership</t>
<t>Suggest to namespace custom properties</t>
<t>Clarify how to handle property record with wrong type</t>
<t>Cover the case of multiple different &lt;unique-N&gt;'s having the same value</t>
<t>Recommendations for naming catalog zones</t>
<t>Add and operational note about notifies for not yet existing zones</t>
<t>Add text about name server restarts with broken zones</t>
<t>Great power comes with great responsibility (Thanks Klaus!)</t>
<t>Mention the new BIND implementation</t>
<t>All invalid properties cause a broken catalog zone, including invalid <tt>group</tt> and <tt>version</tt> properties.</t>
<t>Add Aram Sargsyan as author (he did the BIND9 implementation)</t>
<t><tt>group</tt> properties can have more than one value</t>
</blockquote>
<ul>
<li>draft-toorop-dnsop-dns-catalog-zones-07</li>
</ul>
<blockquote><t>Some spelling fixes from Tim Wicinski and Josh Soref</t>
<t>Replace SHOULDs with MUSTs for ignoring things that are meaningless to a catalog consumer (Thanks Michael StJohns)</t>
<t>Update the list of people to thank in the Acknowledgements section</t>
<t>Mention PowerDNS support of catalog zones from version 4.7.0 onwards</t>
</blockquote>
<ul>
<li>draft-toorop-dnsop-dns-catalog-zones-08</li>
</ul>
<blockquote><t>Address AD Review comments (editorial only)</t>
<t>When DoT is mentioned, also mention now-standardized DoQ</t>
</blockquote>
<ul>
<li>draft-toorop-dnsop-dns-catalog-zones-08</li>
</ul>
<blockquote><t>Editorial nits from David Blacka, Lars Eggert, Russ Housley, Erik Kline, <u format="char-num">É</u>ric Vyncke and Paul Wouters</t>
<t>Addes a Catalog Zone Exampla</t>
<t>Mention that the document uses DNS specific terminology and reference RFC8499</t>
<t>Added IANA Considerations sections, with a registry for Catalog Zones properties</t>
<t>Updated Implementation status also with respect to Catalog zones version &quot;1&quot; support</t>
<t>Updates to Rename &quot;group properties&quot; to &quot;group property values&quot; or &quot;group values&quot; to reduce confusion about who will determine those values (operators and not implementations)</t>
<t>Change example group values in non descriptive names</t>
<t>Add some more clarifications on that and how group values are determined in producer/consumer agreements</t>
<t>Stronger checking suggestion (SHOULD instead of MAY) in accepting member zones by consumers in the Security section</t>
<t>Added mistake recovery text to the Member zone removal section</t>
<t>Replace vague language (&quot;meaningless&quot;) with more precise wording</t>
<t>Catalog consumers that know only version &quot;2&quot; MUST not process version &quot;1&quot; catalog zones and consider it broken.</t>
<t>The entire RDATA of a group property is it's value</t>
</blockquote></section>

</back>

</rfc>
