RFC 10016

System-Defined Configuration

πŸ“… published September 2026, from ietf

System-Defined Configuration πŸŽ€

okay bestie, this is RFC 10016, and it's all about System-Defined Configuration πŸ¦‹ It's a Standards Track doc from the IETF (Internet Engineering Task Force), dated September 2026, with ISSN 2070-1721. The authors are Q. Ma (the editor, "Ed."), Q. Wu, and C. Feng, all from Huawei πŸ’… It updates RFC 8342, and the header doesn't list any Obsoletes entry ✨

Abstract 🌸

So NMDA (the Network Management Datastore Architecture, from RFC 8342) gives us a bunch of configuration datastores holding configuration, and clients are the ones in charge of what goes in them. This doc introduces a system configuration datastore, which holds configuration controlled by the system the server is running on. Clients can reference that system configuration (like with a leafref) from config they create explicitly themselves 🫢

This document updates RFC 8342.

Status of This Memo πŸ“œ

This is an Internet Standards Track document.

This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 7841.

Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at https://www.rfc-editor.org/info/rfc10016.

Copyright (c) 2026 IETF Trust and the persons identified as the document authors. All rights reserved.

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (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 πŸ—‚οΈ

1.  Introduction
  1.1.  Terminology
  1.2.  Requirements Language
  1.3.  Updates to RFC 8342
2.  Kinds of System Configuration
  2.1.  Always Present
  2.2.  Conditionally Present
3.  The System Configuration Datastore (<system>)
4.  Conceptual Model of Datastores
5.  Static Characteristics
  5.1.  Read-Only to Clients
  5.2.  No Changes to <operational>
6.  Dynamic Behaviors
  6.1.  May Change via Software Upgrades or Resource Changes
  6.2.  Referencing System Configuration
  6.3.  Overriding System Configuration
  6.4.  Configuring Descendant Nodes of System Configuration
7.  The "ietf-system-datastore" Module
  7.1.  Data Model Overview
  7.2.  YANG Module
  7.3.  Example Usage
8.  IANA Considerations
  8.1.  The IETF XML Registry
  8.2.  The YANG Module Names Registry
9.  Operational Considerations
10. Security Considerations
  10.1.  Considerations for the "ietf-system-datastore" YANG Module
  10.2.  Considerations for System Configuration
11. References
  11.1.  Normative References
  11.2.  Informative References
Appendix A.  Example of Dynamic Behaviors (Informative)
  A.1.  Referencing System-Defined Nodes
  A.2.  Modifying a System-Instantiated Leaf's Value
  A.3.  Configuring Descendant Nodes of a System-Defined Node
Appendix B.  Key Use Cases (Informative)
  B.1.  Device Powers On
  B.2.  Client Commits Configuration
  B.3.  Operator Installs Card into a Chassis
  B.4.  Client Further Commits Configuration
Acknowledgements
Contributors
Authors' Addresses

1. Introduction ✨

NMDA [RFC8342] defines system configuration as the configuration supplied by the device itself, which appears in <operational> when it is in use (see Figure 2 in [RFC8342]) πŸ’…

But operators want servers to expose system configuration better, even when it isn't in use. Some implementations define system configuration that has to be referenced before it becomes active, for example πŸ‘€ NETCONF / RESTCONF clients would love a standard way to find out what system configuration a server has available 🫢

Some servers also let you configure or modify the descendant nodes of system-defined configuration. Like, the system configuration might include an almost-empty physical interface whose existence is tied to particular hardware, while the client needs to add, modify, or remove a bunch of its descendant nodes. Some descendant nodes might not be modifiable at all (for example, the interface "type" the system sets) πŸ”’

This document updates the NMDA defined in [RFC8342] with a read-only conventional configuration datastore called "system" to expose system-defined configuration. The solution lets configuration created by clients reference nodes defined in <system>, override system-provided values, and configure descendant nodes of system-defined configuration. Main character energy, honestly πŸ’–

The solution defined in this document requires the use of NMDA for both clients and servers. Conformance to this document requires NMDA servers implement the "ietf-system-datastore" YANG module (Section 7) 🌈

1.1. Terminology

This document assumes you already know the contents of [RFC6241], [RFC7950], [RFC8342], and [RFC8525], and it uses terminology from those documents. The terms "device" and "server" are used interchangeably here 🧚

The following terms are defined in this document:

system configuration: [RFC8342] defines it as "Configuration that is supplied by the device itself". This document refines that to mean configuration present in the system configuration datastore (regardless of whether it is applied or referenced). It may also be referred to as "system-defined configuration" or "system-provided configuration" throughout this document. The system configuration discussed in this document cannot be deletable; configuration provided by the server that is deletable is outside the scope of this document 🚫

system configuration datastore: This is a configuration datastore holding configuration provided by the system itself. This datastore is referred to as <system> πŸ’œ

This document redefines the term "conventional configuration datastore" in Section 3 of [RFC8342] to add "system" to the list of conventional configuration datastores:

conventional configuration datastore: One of the following set of configuration datastores: <running>, <startup>, <candidate>, <system>, and <intended>. These datastores share a common datastore schema, and protocol operations allow copying data between them. The term "conventional" is a generic umbrella term for these datastores. Note that while protocol operations allow copying data between conventional datastores, the read-only nature of datastores such as <system> and <intended> restricts clients from copying data into them πŸ“‹

system node: This is an instance in the data tree that is provided by the system itself. A system node may also be called a "system-defined node" or "system-provided node" throughout this document 🌸

referenced node: A referenced node is one of the following:

  • Targets of leafref values defined via the path statement.
  • Targets of instance-identifier type values.
  • Nodes present in an XPath expression of when constraints.
  • Nodes present in an XPath expression of must constraints.
  • Nodes defined to satisfy the mandatory true constraints.
  • Nodes defined to satisfy the min-elements constraints.

1.2. Requirements Language

Here's the fine print on the key words. 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 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here πŸŽ€

1.3. Updates to RFC 8342

This document updates [RFC8342] to define a configuration datastore called "system" that holds system configuration (Section 3). It also redefines the term "conventional configuration datastore" from [RFC8342] to include "system" in the list πŸ¦‹

To keep <intended> valid when clients interact with system configuration (for example, when configuration provided by clients references system configuration), and to allow system-defined templates and inactive system configuration, the configuration in <running> is merged with <system> to create the contents of <intended>. This happens after the configuration transformations (e.g., template expansion or removal of inactive configuration defined in [RFC8342]) have been performed, as described in Section 4. The specifications in [RFC8342] related to processing system configuration are updated by the mechanism defined in this document βš™οΈ

Additionally, this document updates the definition of the "intended" origin metadata annotation identity defined in Section 5.3.4 of [RFC8342]. The "intended" identity of the origin value defined in [RFC8342] represents the origin of configuration provided by <intended>. This document updates that definition so it is the origin source of configuration explicitly provided by clients. It also allows a subset of configuration in <intended> that flows from <system>, yet is not configured or overridden explicitly in <running>, to use "system" as its origin value. As per Section 5.3.4 of [RFC8342], all configuration with the origin value being reported as "intended" MUST originate from <running>, which includes any configuration in <system> that has been copied into <running>. Configuration that is in <system> and not also present in <running> MUST be reported as origin "system" in <operational> πŸ”₯

2. Kinds of System Configuration 🧚

This document defines two types of system configuration: configuration that is always present, and configuration that is conditionally present. These types are described in Sections 2.1 and 2.2, respectively πŸ’…

2.1. Always Present

The always-present system configuration is generated in <system> when the device is powered on, irrespective of whether physical resources are present or whether a special functionality is enabled. An example of always-present system configuration is an always-existing loopback interface 🌊

2.2. Conditionally Present

The conditionally present system configuration is generated in <system> based on specific conditions being met in a system. For example, if a physical resource is present (e.g., an interface card is inserted), the system automatically detects it and loads the associated configuration. When the physical resource is not present (an interface card is removed), the system configuration will automatically be removed from <system>. Another example is when a special functionality (e.g., a license or feature) is enabled, in which case specific configuration may be created by the system πŸ“

3. The System Configuration Datastore (<system>) ✨

Okay so, following the datastore rules in Appendix A of [RFC8342], this doc introduces a brand new datastore called "system" πŸŽ€ It's where the system configuration lives. NMDA servers that follow this doc MUST implement a system configuration datastore, and they SHOULD also implement <intended> πŸ’…

  • Name: "system"
  • YANG modules: all
  • YANG nodes: all "config true" data nodes up to the root node, generated by the system.
  • Management operations: you can read the datastore with network management protocols like NETCONF and RESTCONF, but you can't change its contents through NETCONF or RESTCONF. Not me trying to edit it πŸ™ƒ
  • Origin: this doc doesn't define any new origin identity. Instead, the "system" identity of the origin metadata annotation [RFC7952] shows where a data item in <system> came from 🧚
  • Protocols: YANG-driven management protocols, such as NETCONF and RESTCONF.
  • Defining YANG module: "ietf-system-datastore" (Section 7).

The system configuration datastore does not persist across reboots πŸ’€ So it's giving temporary, but make it iconic.

4. Conceptual Model of Datastores 🌈

Clients can provide configuration nodes that point at nodes defined in <system>, override the values the system set, and configure descendant nodes of system-defined configuration in <running>. You'll see the details in Section 6 🫢

Now, to keep <intended> valid, the configuration in <running> gets merged with <system> to become <intended>. When they clash, whatever is in <running> wins over the same node in <system> πŸ‘‘ Here's the tricky part: the doc says it's unspecified how to merge configuration before transformations. So if <system> or <running> has stuff that needs more transformation first (like template expansion or removing inactive configuration defined in [RFC8342]) before it can be applied, those transformations MUST be done separately on each datastore before <running> gets merged with <system>. Order matters, bestie 😭

Whenever the configuration in <system> changes, the server MUST also immediately update and validate <intended> ✨

Okay, so Figure 2 in Section 5 of [RFC8342] gets updated with the conceptual model below, which includes the system configuration datastore. For completeness, this model also includes <factory-default>, introduced in [RFC8808] πŸ¦‹

                              +-----------------+
                              |<factory-default>|
                        +-----|    (ct, ro)     |-----+
                        |     +-----------------+     |
   //"factory-reset" RPC|             |               |
                        v             |               v
                +-------------+       |         +-----------+
                | <candidate> |       |         | <startup> |
                |  (ct, rw)   |<---+  |   +---->| (ct, rw)  |
                +-------------+    |  |   |     +-----------+
                       |           |  v   |           |
 +-----------+         |        +-----------+         |
 | <system>  |         +------->| <running> |<--------+
 | (ct, ro)  |                  | (ct, rw)  |
 +-----------+                  +-----------+
      |                              |
      |                              |
      |                              | // configuration transformations,
      +--------------+---------------+ // e.g., removal of nodes marked
                     |                 // as "inactive", expansion of
                     |                 // templates
                     v
               +------------+
               | <intended> |  // subject to validation
               | (ct, ro)   |
               +------------+
                      |       // changes applied, subject to
                      |       // local factors, e.g., missing
                      |       // resources, delays
  dynamic             |
  configuration       |   +-------- learned configuration
  datastores -----+   |   +-------- default configuration
                  |   |   |
                  v   v   v
              +---------------+
              | <operational> | <-- system state
              | (ct + cf, ro) |
              +---------------+

 ct = config true; cf = config false
 rw = read-write; ro = read-only
 boxes denote named datastores

Figure 1: Architectural Model of Datastores πŸ“

Configuration in <system> can't be deleted by clients. A list entry can never be removed from <system> through protocol operations πŸ™…β€β™€οΈ Even so, a node defined in <system> can still be overridden in <running>. Here's the vibe: if the system sets an initial value for a leaf, and the client overrides it with a different value in <running> (Section 6.3), and then that node in <running> gets removed later, the system-initialized value from <system> shows up in <intended> and may come into use eventually if it's applied successfully πŸ”„

Configuration may also disappear from <system> because of things like resources that are no longer available πŸ₯€ When that happens, the configuration for missing resources can still hang out in <running> and <intended>, but it won't be applied or appear in <operational>. Section 5.3.2 of [RFC8342] explains this in more detail 🧐

5. Static Characteristics πŸŽ€

5.1. Read-Only to Clients

The system datastore is read-only, so edits aimed directly at <system> MUST be denied. Clients might still be allowed to provide configuration that overrides the value of a system-initialized node, though (see Section 6.3) 🫧

5.2. No Changes to <operational>

This work doesn't change the definition of <operational>, and it doesn't affect what's in <operational>, as specified in [RFC8342] πŸ™Œ What it does do is clarify origin reporting. So the origin of nodes sourced from <system> is reported as "system" unless they're explicitly configured or overridden in <running> 🏷️ <system> lets system-defined nodes be defined like configuration, meaning they're visible to clients so they can be referenced or configured before they show up in <operational>. "config false" nodes are out of scope, so existing "config false" nodes aren't affected by this work 🀍

6. Dynamic Behaviors 🌸

6.1. May Change via Software Upgrades or Resource Changes

The contents of <system> MAY change dynamically under various conditions, like a license change, a software upgrade, or a system-controlled resource change (see Section 2.2) πŸ” Updates to the system configuration can be obtained through YANG notifications [RFC8639] [RFC8641] (for example, an on-change notification) πŸ“¬

If the system configuration changes (say, during a software upgrade), <running> SHOULD remain a valid configuration data tree ✨ Any mechanisms for making that happen are outside the scope of this document, so this doc isn't going to spill those secrets 🀐

6.2. Referencing System Configuration

Clients can create configuration data in <running> that references nodes in <system> πŸ”— Some implementations may define system nodes purely as a convenience for clients to reference. Clients can also define their own customized nodes to reference, which is totally a thing πŸ’…

Appendix A.1 has an example of a client referencing system-defined nodes πŸ“–

6.3. Overriding System Configuration

Even though <system> is read-only, a server may allow some parts of the system configuration (like a leaf's value) to be overridden. Quick note: <system> and the system configuration are not the same thing, so keep that distinction in mind 🧠 Overriding works by the client writing configuration data in <running>, which overrides the values of matched configuration nodes at the matching level in <system>. Configurations in <running> take precedence over system configuration nodes in <system> if the server allows those nodes to be overridden. Some implementations may have immutable system configuration, which the server identifies with an immutable metadata annotation (see [YANG-FLAG] for details). Overrides win regardless of whether a system-instantiated value changes later πŸ†

Appendix A.2 has an example of a client overriding a system-instantiated leaf's value 🫢

6.4. Configuring Descendant Nodes of System Configuration

A server may also let a client add nodes to a list entry in <system> by writing those extra nodes in <running> 🌱 These additional data nodes may not exist in <system> at all, so this is an addition rather than an override πŸ“

Appendix A.3 has an example of a client configuring descendant nodes of a system-defined node 🧁

7. The "ietf-system-datastore" Module πŸ’œ

7.1. Data Model Overview

This YANG module defines a new YANG identity named "system". It uses the "ds:conventional" identity defined in [RFC8342] as its base 🧩 A client can discover whether the server supports the system configuration datastore by reading the YANG library information from the operational state datastore πŸ”

The system datastore is defined as a conventional configuration datastore, so it shares a common datastore schema with the other conventional datastores 🀝

The diagram below shows how the "identity" statements in the "ietf-system-datastore" and "ietf-datastores" YANG modules relate to each other πŸ¦‹

Identities:
  +--- datastore
  |  +--- conventional
  |  |  +--- running
  |  |  +--- candidate
  |  |  +--- startup
  |  |  +--- system
  |  |  +--- intended
  |  +--- dynamic
  |  +--- operational

The diagram above uses syntax that is similar to, but not defined in, [RFC8340] πŸ€“

7.2. YANG Module

Here's the YANG module, verbatim πŸ’–

<CODE BEGINS> file "ietf-system-datastore@2026-09-16.yang"
module ietf-system-datastore {
  yang-version 1.1;
  namespace "urn:ietf:params:xml:ns:yang:ietf-system-datastore";
  prefix sysds;

  import ietf-datastores {
    prefix ds;
    reference
      "RFC 8342: Network Management Datastore Architecture (NMDA)";
  }

  organization
    "IETF NETMOD (Network Modeling) Working Group";

  contact
    "WG Web:   <https://datatracker.ietf.org/wg/netmod/>
     WG List:  <mailto:netmod@ietf.org>

     Author: Qiufang Ma
             <mailto:maqiufang1@huawei.com>
     Author: Qin Wu
             <mailto:bill.wu@huawei.com>
     Author: Chong Feng
             <mailto:fengchongllly@gmail.com>";

  description
    "This module defines a new YANG identity that uses the
     ds:conventional identity defined in RFC 8342.

     Copyright (c) 2026 IETF Trust and the persons identified
     as authors of the code.  All rights reserved.

     Redistribution and use in source and binary forms, with
     or without modification, is permitted pursuant to, and
     subject to the license terms contained in, the Revised
     BSD License set forth in Section 4.c of the IETF Trust's
     Legal Provisions Relating to IETF Documents
     (https://trustee.ietf.org/license-info).

     This version of this YANG module is part of RFC 10016
     (https://www.rfc-editor.org/info/rfc10016); see the RFC
     itself for full legal notices.";

  revision 2026-09-16 {
    description
      "Initial version.";
    reference
      "RFC 10016: System-Defined Configuration";
  }

  identity system {
    base ds:conventional;
    description
      "This read-only datastore contains the configuration
       provided by the system itself.";
  }
}
<CODE ENDS>

7.3. Example Usage ✨

Okay, let's see this in action πŸŽ€ The example below shows how the configuration in <system> can be pulled back with a NETCONF <get-data> RPC operation. It uses the "example-application" fictional data model from Appendix A.1 and the Extensible Markup Language (XML) [W3C.XML1.0] encoding 🧚

   <rpc message-id="101"
        xmlns="urn:ietf:params:xml:ns:netconf:base:1.0">
     <get-data xmlns="urn:ietf:params:yang:ietf-netconf-nmda"
        xmlns:sysds="urn:ietf:params:xml:ns:yang:ietf-system-datastore">
       <datastore>sysds:system</datastore>
       <subtree-filter>
         <applications xmlns="urn:example:application"/>
       </subtree-filter>
     </get-data>
   </rpc>

Now for RESTCONF 🌸 the system configuration datastore can be reached at the resource {+restconf}/ds/ietf-system-datastore:system. Here's an HTTP GET method asking for the "applications" configuration:

   GET /restconf/ds/ietf-system-datastore:system/\
      example-application:applications HTTP/1.1
   Host: example.com
   Accept: application/yang-data+xml

8. IANA Considerations πŸ’…

8.1. The IETF XML Registry

IANA has registered the following XML namespace URI in the "ns" registry within the "IETF XML Registry" group [RFC3688] πŸ“‡

   URI:  urn:ietf:params:xml:ns:yang:ietf-system-datastore
   Registrant Contact:  The IESG
   XML:  N/A; the requested URIs are XML namespaces.

8.2. The YANG Module Names Registry

IANA has also registered the following YANG module in the "YANG Module Names" registry within the "YANG Parameters" registry group. [RFC6020].

   Name:  ietf-system-datastore
   Maintained by IANA?  N
   Namespace:  urn:ietf:params:xml:ns:yang:ietf-system-datastore
   Prefix:  sysds
   Reference:  RFC 10016

9. Operational Considerations 🫢

Okay, so system configuration exists whether or not a server implements <system> 🌈 The introduction of <system> gives us a standardized way to expose system configuration within NMDA, and that's honestly iconic.

NMDA clients that are not aware of <system> will keep working just fine πŸ’– They'll only interact with datastores such as <running>, <candidate>, <intended>, and <operational>, same as before. The presence of <system> doesn't change the fundamental behavior for those legacy clients. Operators should be aware that to fully leverage the capabilities in this document, client applications need to be updated to recognize and interact with <system> ✨

10. Security Considerations πŸ”

10.1. Considerations for the "ietf-system-datastore" YANG Module

This section is modeled after the template described in Section 3.7.1 of [RFC9907], so it's very organized πŸ“‹

The "ietf-system-datastore" YANG module is designed to be accessed via YANG-based management protocols, such as the Network Configuration Protocol (NETCONF) [RFC6241] and RESTCONF [RFC8040]. These YANG-based management protocols (1) have to use a secure transport layer and (2) have to use mutual authentication (e.g., Secure Shell (SSH) [RFC4252], TLS [TLS1.3], and QUIC [RFC9000]) πŸ”’

The Network Configuration Access Control Model (NACM) [RFC8341] provides the means to restrict access for particular NETCONF or RESTCONF users to a preconfigured subset of all available NETCONF or RESTCONF protocol operations and content πŸ›‘οΈ

The YANG module only defines an identity that uses the ds:conventional identity as its base. The module by itself does not expose any sensitive data nodes that are writable or that contain read-only state, and there are no particularly sensitive RPC or action operations. As such, there are no additional security issues related to the YANG module that need to be considered πŸ™ˆ

10.2. Considerations for System Configuration

Okay, real talk 🚨 The system datastore is read-only to clients, but it can still hold sensitive stuff like hardware identifiers, security policies, and critical system resources. Read access to sensitive system nodes and subtrees within the datastore MUST be controlled to prevent unauthorized disclosure. Implementations are strongly advised to log all access attempts to sensitive system configuration for audit purposes πŸ•΅οΈ

Plot twist 😱 even though <system> cannot be modified directly, system configuration can still be overridden as a merging result (Section 6.3). An attacker may configure a leaf that shadows a sensitive node in <system>. Misconfiguration in <running> could lead to unintended system behavior, including security policy bypass and availability risks. Unauthorized modification to sensitive contents MUST be prevented to avoid those negative effects on the network πŸ’€

11. References πŸ“š

11.1. Normative References

11.2. Informative References

Appendix A. Example of Dynamic Behaviors (Informative) 🧚

This section presents some sample data models and the matching contents of various datastores with different dynamic behaviors described in Section 6, which is the fun part 🎠 The XML snippets are only for illustration purposes. Note that this section does not show the contents of <intended>, since those are related to the configuration in <operational>, assuming the intended configuration is applied successfully. Also note that if the "origin" metadata annotation for configuration is unspecified in snippets, it is inherited from its parent node 🫢

A.1. Referencing System-Defined Nodes ✨

okay bestie, for this bit we're using a made-up module called example-application 🧚. Nothing real, just vibes:

module example-application {
  yang-version 1.1;
  namespace "urn:example:application";
  prefix ex-app;

  import ietf-inet-types {
    prefix inet;
  }
  container applications {
    list application {
      key "name";
      leaf name {
        type string;
      }
      leaf app-id {
        type string;
      }
      leaf protocol {
        type enumeration {
          enum tcp;
          enum udp;
        }
        mandatory true;
      }
      leaf destination-port {
        default "0";
        type inet:port-number;
      }
      leaf description {
        type string;
      }
      container security-protection {
        presence "Indicates that security protection is enabled.";
        leaf risk-level {
          type enumeration {
            enum high;
            enum low;
          }
        }
        //additional leafs for security-specific configuration...
      }
    }
  }
}

Next up is a fictional Access Control List (ACL) module πŸ’…. Its big moment is a leafref on the leaf-list "application" that points at an application name that already exists. It's basically the ACL going "hey, this rule is talking about ftp, go find the real ftp entry" πŸ”₯.

module example-acl {
  yang-version 1.1;
  namespace "urn:example:acl";
  prefix ex-acl;

  import example-application {
    prefix ex-app;
  }

  import ietf-inet-types {
    prefix inet;
  }

  container acl {
    list acl-rule {
      key "name";
      leaf name {
        type string;
      }
      container matches {
        choice l3 {
          container ipv4 {
            leaf src-address {
              type inet:ipv4-prefix;
            }
            leaf dst-address {
              type inet:ipv4-prefix;
            }
          }
        }
        choice applications {
          leaf-list application {
            type leafref {
              path "/ex-app:applications/ex-app:application"
                 + "/ex-app:name";
            }
          }
        }
      }
      leaf packet-action {
        type enumeration {
          enum forward;
          enum drop;
          enum redirect;
        }
      }
    }
  }
}

The server can pre-define some applications as a little gift for clients πŸŽ€. These are immediately there in system configuration, and when the device powers on, the system-instantiated application entries may show up in <system> like this:

<applications xmlns="urn:example:application">
  <application>
    <name>ftp</name>
    <app-id>001</app-id>
    <protocol>tcp</protocol>
    <destination-port>21</destination-port>
    <security-protection>
      <risk-level>low</risk-level>
    </security-protection>
  </application>
  <application>
    <name>tftp</name>
    <app-id>002</app-id>
    <protocol>udp</protocol>
    <destination-port>69</destination-port>
    <security-protection>
      <risk-level>low</risk-level>
    </security-protection>
  </application>
  <application>
    <name>smtp</name>
    <app-id>003</app-id>
    <protocol>tcp</protocol>
    <destination-port>25</destination-port>
    <security-protection>
      <risk-level>low</risk-level>
    </security-protection>
  </application>
</applications>

Clients can also make their own customized applications 🌸. Those may show up in <running> like so:

<applications xmlns="urn:example:application">
  <application>
    <name>my-smtp</name>
    <app-id>101</app-id>
    <protocol>tcp</protocol>
    <destination-port>2345</destination-port>
    <description>customized smtp application</description>
    <security-protection>
      <risk-level>high</risk-level>
    </security-protection>
  </application>
  <application>
    <name>my-foo</name>
    <app-id>102</app-id>
    <protocol>udp</protocol>
    <destination-port>1024</destination-port>
    <description>customized application</description>
  </application>
</applications>

If a client sets up an ACL rule that references some system-provided or customized applications, the rule may look like this πŸ¦‹:

<acl xmlns="urn:example:acl">
  <acl-rule>
    <name>allow-access-to-ftp-tftp</name>
    <matches>
      <ipv4>
        <src-address>198.51.100.0/24</src-address>
        <dst-address>192.0.2.0/24</dst-address>
      </ipv4>
      <application>ftp</application>
      <application>tftp</application>
      <application>my-smtp</application>
    </matches>
    <packet-action>forward</packet-action>
  </acl-rule>
</acl>

Now here's the glow-up πŸ‘‘. The application entries from <system> and <running> get merged to create <intended>. There are no merging conflicts in the contents between <system> and <running>, so <operational> might show the application config with origin values that reflect where each entry came from:

<applications xmlns="urn:example:application"
              xmlns:or="urn:ietf:params:xml:ns:yang:ietf-origin"
              or:origin="or:intended">
  <application>
    <name>my-smtp</name>
    <app-id>101</app-id>
    <protocol>tcp</protocol>
    <destination-port>2345</destination-port>
    <description>customized smtp application</description>
    <security-protection>
      <risk-level>high</risk-level>
    </security-protection>
  </application>
  <application>
    <name>my-foo</name>
    <app-id>102</app-id>
    <protocol>udp</protocol>
    <destination-port>1024</destination-port>
    <description>customized application</description>
  </application>
  <application or:origin="or:system">
    <name>ftp</name>
    <app-id>001</app-id>
    <protocol>tcp</protocol>
    <destination-port>21</destination-port>
    <security-protection>
      <risk-level>low</risk-level>
    </security-protection>
  </application>
  <application or:origin="or:system">
    <name>tftp</name>
    <app-id>002</app-id>
    <protocol>udp</protocol>
    <destination-port>69</destination-port>
    <security-protection>
      <risk-level>low</risk-level>
    </security-protection>
  </application>
  <application or:origin="or:system">
    <name>smtp</name>
    <app-id>003</app-id>
    <protocol>tcp</protocol>
    <destination-port>25</destination-port>
    <security-protection>
      <risk-level>low</risk-level>
    </security-protection>
  </application>
</applications>

A.2. Modifying a System-Instantiated Leaf's Value πŸ¦‹

This one uses another fictional interface module, and it's giving "system-provided defaults with a client twist" 🌈:

module example-interface {
  yang-version 1.1;
  namespace "urn:example:interface";
  prefix ex-if;

  import ietf-inet-types {
    prefix inet;
  }

  container interfaces {
    list interface {
      key "name";
      leaf name {
        type string;
      }
      leaf description {
        type string;
      }
      leaf mtu {
        type uint32;
      }
      leaf-list ip-address {
        type inet:ip-address;
      }
    }
  }
}

Say the system always has a loopback interface called "lo0" πŸ’–. It comes with an MTU value of "65536", a default IPv4 address of "127.0.0.1", and a default IPv6 address of "::1". That configuration may be present in <system> as follows:

<interfaces xmlns="urn:example:interface">
  <interface>
    <name>lo0</name>
    <mtu>65536</mtu>
    <ip-address>127.0.0.1</ip-address>
    <ip-address>::1</ip-address>
  </interface>
</interfaces>

Then a client changes the MTU to 9216 by adding this to <running> πŸ’…:

<interfaces xmlns="urn:example:interface">
  <interface>
    <name>lo0</name>
    <mtu>9216</mtu>
  </interface>
</interfaces>

The client's MTU wins over the system-provided one, and since the "origin" value of the client's configuration is set to "intended", the interfaces configuration in <operational> may be as follows πŸ”₯:

<interfaces xmlns="urn:example:interface"
            xmlns:or="urn:ietf:params:xml:ns:yang:ietf-origin"
            or:origin="or:intended">
  <interface>
    <name>lo0</name>
    <mtu>9216</mtu>
    <ip-address or:origin="or:system">127.0.0.1</ip-address>
    <ip-address or:origin="or:system">::1</ip-address>
  </interface>
</interfaces>

A.3. Configuring Descendant Nodes of a System-Defined Node ✨

Picking up from Appendix A.2 🧚, imagine the client also adds a description to the "lo0" interface in <running>:

<interfaces xmlns="urn:example:interface">
  <interface>
    <name>lo0</name>
    <description>loopback</description>
  </interface>
</interfaces>

The configuration of interface "lo0" then shows up in <operational> like this πŸ’œ:

<interfaces xmlns="urn:example:interface"
            xmlns:or="urn:ietf:params:xml:ns:yang:ietf-origin"
            or:origin="or:intended">
  <interface>
    <name>lo0</name>
    <description>loopback</description>
    <mtu>9216</mtu>
    <ip-address or:origin="or:system">127.0.0.1</ip-address>
    <ip-address or:origin="or:system">::1</ip-address>
  </interface>
</interfaces>

Appendix B. Key Use Cases (Informative) πŸ“

This section updates the "Interface Example" supplied in Appendix C.3 of [RFC8342].

This section covers several use cases for how <system> interacts with the other datastores (like <candidate>, <running>, <intended>, and <operational>) 🫢. We're using one more fictional interface data model:

module example-interface-management {
  yang-version 1.1;
  namespace "urn:example:interfacemgmt";
  prefix ex-ifm;

  import ietf-inet-types {
    prefix inet;
  }

  container interfaces {
    list interface {
      key "name";
      leaf name {
        type string;
      }
      leaf type {
        type enumeration {
          enum ethernet;
          enum atm;
          enum loopback;
        }
      }
      leaf enabled {
        type boolean;
        default "true";
      }
      leaf-list ip-address {
        type inet:ip-address;
      }
      leaf speed {
        when "../type = 'ethernet'";
        type enumeration {
          enum 10Mb;
          enum 100Mb;
        }
      }
      leaf description {
        type string;
      }
    }
  }
}

For each use case, you'll get sample configuration in <running>, <system>, <intended>, and <operational> πŸ’–. The XML snippets are only for illustration purposes, so keep that in mind as we go 🌸.

B.1. Device Powers On 🌸

okay so picture this: the device powers on, and we're just gonna assume the system gives you an always-present loopback interface called "lo0" πŸ¦‹. nobody had to configure it, so there's zero interface config in <running>. <system> ends up looking like this:

<interfaces xmlns="urn:example:interfacemgmt">
  <interface>
    <name>lo0</name>
    <type>loopback</type>
    <ip-address>127.0.0.1</ip-address>
    <ip-address>::1</ip-address>
    <description>system-defined interface</description>
  </interface>
</interfaces>

lowkey the loopback config only lives in <system>, and the config for the interface in <intended> is identical to the one in <system> above πŸ’…

<operational> shows the system-provided loopback too! and it also includes the default value from the YANG module, which is that enabled true bit πŸ‘‘

<interfaces xmlns="urn:example:interfacemgmt"
            xmlns:or="urn:ietf:params:xml:ns:yang:ietf-origin"
            or:origin="or:system">
  <interface>
    <name>lo0</name>
    <type>loopback</type>
    <enabled or:origin="or:default">true</enabled>
    <ip-address>127.0.0.1</ip-address>
    <ip-address>::1</ip-address>
    <description>system-defined interface</description>
  </interface>
</interfaces>

B.2. Client Commits Configuration πŸ’–

so a client creates an interface "et-0/0/0", but it's not physically there yet πŸ™ˆ <running> looks like this:

<interfaces xmlns="urn:example:interfacemgmt">
  <interface>
    <name>et-0/0/0</name>
    <ip-address>192.168.10.10</ip-address>
    <description>pre-provisioned interface</description>
  </interface>
</interfaces>

and <system> doesn't change at all. it's still just "lo0", since "et-0/0/0" isn't physically present:

<interfaces xmlns="urn:example:interfacemgmt">
  <interface>
    <name>lo0</name>
    <type>loopback</type>
    <ip-address>127.0.0.1</ip-address>
    <ip-address>::1</ip-address>
    <description>system-defined interface</description>
  </interface>
</interfaces>

<intended> is the merged data of <system> and <running> 🧠

<interfaces xmlns="urn:example:interfacemgmt">
  <interface>
    <name>lo0</name>
    <type>loopback</type>
    <ip-address>127.0.0.1</ip-address>
    <ip-address>::1</ip-address>
    <description>system-defined interface</description>
  </interface>
  <interface>
    <name>et-0/0/0</name>
    <ip-address>192.168.10.10</ip-address>
    <description>pre-provisioned interface</description>
  </interface>
</interfaces>

since "et-0/0/0" doesn't exist, its config isn't in <operational>. it looks like this instead πŸ₯Ή

<interfaces xmlns="urn:example:interfacemgmt"
            xmlns:or="urn:ietf:params:xml:ns:yang:ietf-origin"
            or:origin="or:intended">
  <interface or:origin="or:system">
    <name>lo0</name>
    <type>loopback</type>
    <enabled or:origin="or:default">true</enabled>
    <ip-address>127.0.0.1</ip-address>
    <ip-address>::1</ip-address>
    <description>system-defined interface</description>
  </interface>
</interfaces>

B.3. Operator Installs Card into a Chassis πŸ”₯

okay this one is so satisfying. when the operator installs the interface, the system detects it and generates the associated conditionally present interface config in <system> ✨ <running> stays exactly the same:

<interfaces xmlns="urn:example:interfacemgmt">
  <interface>
    <name>et-0/0/0</name>
    <ip-address>192.168.10.10</ip-address>
    <description>pre-provisioned interface</description>
  </interface>
</interfaces>

and <system> might look like this:

<interfaces xmlns="urn:example:interfacemgmt">
  <interface>
    <name>lo0</name>
    <type>loopback</type>
    <ip-address>127.0.0.1</ip-address>
    <ip-address>::1</ip-address>
    <description>system-defined interface</description>
  </interface>
  <interface>
    <name>et-0/0/0</name>
    <type>ethernet</type>
    <description>system-defined interface</description>
  </interface>
</interfaces>

then <intended> has the merged config of <system> and <running>. main character energy πŸ’…

<interfaces xmlns="urn:example:interfacemgmt">
  <interface>
    <name>lo0</name>
    <type>loopback</type>
    <ip-address>127.0.0.1</ip-address>
    <ip-address>::1</ip-address>
    <description>system-defined interface</description>
  </interface>
  <interface>
    <name>et-0/0/0</name>
    <type>ethernet</type>
    <ip-address>192.168.10.10</ip-address>
    <description>pre-provisioned interface</description>
  </interface>
</interfaces>

and <operational> appears as follows πŸ¦‹

<interfaces xmlns="urn:example:interfacemgmt"
            xmlns:or="urn:ietf:params:xml:ns:yang:ietf-origin"
            or:origin="or:intended">
  <interface or:origin="or:system">
    <name>lo0</name>
    <type>loopback</type>
    <enabled or:origin="or:default">true</enabled>
    <ip-address>127.0.0.1</ip-address>
    <ip-address>::1</ip-address>
    <description>system-defined interface</description>
  </interface>
  <interface>
    <name>et-0/0/0</name>
    <type or:origin="or:system">ethernet</type>
    <enabled or:origin="or:default">true</enabled>
    <ip-address>192.168.10.10</ip-address>
    <description>pre-provisioned interface</description>
  </interface>
</interfaces>

B.4. Client Further Commits Configuration 🧚

okay so now the client goes further and sets the speed of "et-0/0/0" in <running> ⚑

<interfaces xmlns="urn:example:interfacemgmt">
  <interface>
    <name>et-0/0/0</name>
    <speed>10Mb</speed>
  </interface>
</interfaces>

<system> stays unchanged 🫢

<interfaces xmlns="urn:example:interfacemgmt">
  <interface>
    <name>lo0</name>
    <type>loopback</type>
    <ip-address>127.0.0.1</ip-address>
    <ip-address>::1</ip-address>
    <description>system-defined interface</description>
  </interface>
  <interface>
    <name>et-0/0/0</name>
    <type>ethernet</type>
    <description>system-defined interface</description>
  </interface>
</interfaces>

and <intended> is the merged result of <running> and <system>, so it looks like this:

<interfaces xmlns="urn:example:interfacemgmt">
  <interface>
    <name>lo0</name>
    <type>loopback</type>
    <ip-address>127.0.0.1</ip-address>
    <ip-address>::1</ip-address>
    <description>system-defined interface</description>
  </interface>
  <interface>
    <name>et-0/0/0</name>
    <type>ethernet</type>
    <ip-address>192.168.10.10</ip-address>
    <speed>10Mb</speed>
    <description>pre-provisioned interface</description>
  </interface>
</interfaces>

and <operational> would look like this, bestie πŸŽ€

<interfaces xmlns="urn:example:interfacemgmt"
           xmlns:or="urn:ietf:params:xml:ns:yang:ietf-origin"
           or:origin="or:intended">
 <interface or:origin="or:system">
   <name>lo0</name>
   <type>loopback</type>
   <enabled or:origin="or:default">true</enabled>
   <ip-address>127.0.0.1</ip-address>
   <ip-address>::1</ip-address>
   <description>system-defined interface</description>
 </interface>
 <interface>
   <name>et-0/0/0</name>
   <type or:origin="or:system">ethernet</type>
   <enabled or:origin="or:default">true</enabled>
   <ip-address>192.168.10.10</ip-address>
   <speed>10Mb</speed>
   <description>pre-provisioned interface</description>
 </interface>
</interfaces>

Acknowledgements 🫢

the authors wanna thank these legends for the discussions and for giving input on this document: BalΓ‘zs Lengyel, Robert Wilton, JΓΌrgen SchΓΆnwΓ€lder, Andy Bierman, Martin BjΓΆrklund, Mohamed Boucadair, Michal VaΕ‘ko, Alexander Clemm, and Timothy Carey πŸ’œ

Contributors 🌈

Kent Watsen Watsen Networks Email: kent+ietf@watsen.net

Jan Lindblad Cisco Systems Email: jlindbla@cisco.com

Jason Sterne Nokia Email: jason.sterne@nokia.com

Chongfeng Xie China Telecom Beijing China Email: xiechf@chinatelecom.cn

Authors' Addresses πŸ“

Qiufang Ma (editor) Huawei 101 Software Avenue, Yuhua District Nanjing Jiangsu, 210012 China Email: maqiufang1@huawei.com

Qin Wu Huawei 101 Software Avenue, Yuhua District Nanjing Jiangsu, 210012 China Email: bill.wu@huawei.com

Chong Feng Email: fengchongllly@gmail.com

πŸ“¦ where this page comes from