--- 1/draft-ietf-netmod-revised-datastores-01.txt 2017-05-11 13:13:20.320094143 -0700 +++ 2/draft-ietf-netmod-revised-datastores-02.txt 2017-05-11 13:13:20.384095676 -0700 @@ -1,24 +1,24 @@ Network Working Group M. Bjorklund Internet-Draft Tail-f Systems Intended status: Standards Track J. Schoenwaelder -Expires: September 14, 2017 Jacobs University +Expires: November 12, 2017 Jacobs University P. Shafer K. Watsen Juniper Networks R. Wilton Cisco Systems - March 13, 2017 + May 11, 2017 Network Management Datastore Architecture - draft-ietf-netmod-revised-datastores-01 + draft-ietf-netmod-revised-datastores-02 Abstract Datastores are a fundamental concept binding the data models written in the YANG data modeling language to network management protocols such as NETCONF and RESTCONF. This document defines an architectural framework for datastores based on the experience gained with the initial simpler model, addressing requirements that were not well supported in the initial model. @@ -30,262 +30,258 @@ Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http://datatracker.ietf.org/drafts/current/. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." - This Internet-Draft will expire on September 14, 2017. + This Internet-Draft will expire on November 12, 2017. Copyright Notice Copyright (c) 2017 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3 - 3. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 5 - 3.1. Original Model of Datastores . . . . . . . . . . . . . . 7 + 3. Background . . . . . . . . . . . . . . . . . . . . . . . . . 5 + 3.1. Original Model of Datastores . . . . . . . . . . . . . . 6 4. Architectural Model of Datastores . . . . . . . . . . . . . . 8 - 4.1. The Datastore . . . . . . . . . . . . . . . . 9 - 4.2. Dynamic Datastores . . . . . . . . . . . . . . . . . . . 10 - 4.3. The Datastore . . . . . . . . . . . . . . . 10 - 4.3.1. Missing Resources . . . . . . . . . . . . . . . . . . 11 - 4.3.2. System-controlled Resources . . . . . . . . . . . . . 11 - 4.3.3. Origin Metadata Annotation . . . . . . . . . . . . . 11 - 5. Guidelines for Defining Dynamic Datastores . . . . . . . . . 12 - 5.1. Define a name for the dynamic datastore . . . . . . . . . 12 - 5.2. Define which YANG modules can be used in the datastore . 12 - 5.3. Define which subset of YANG-modeled data applies . . . . 13 - 5.4. Define how dynamic data is actualized . . . . . . . . . . 13 - 5.5. Define which protocols can be used . . . . . . . . . . . 13 - 5.6. Define a module for the dynamic datastore . . . . . . . . 13 - 6. YANG Modules . . . . . . . . . . . . . . . . . . . . . . . . 14 - 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 18 - 7.1. Updates to the IETF XML Registry . . . . . . . . . . . . 18 - 7.2. Updates to the YANG Module Names Registry . . . . . . . . 19 - 8. Security Considerations . . . . . . . . . . . . . . . . . . . 19 - 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 19 - 10. References . . . . . . . . . . . . . . . . . . . . . . . . . 20 - 10.1. Normative References . . . . . . . . . . . . . . . . . . 20 - 10.2. Informative References . . . . . . . . . . . . . . . . . 21 - Appendix A. Example Data . . . . . . . . . . . . . . . . . . . . 22 - A.1. System Example . . . . . . . . . . . . . . . . . . . . . 22 - A.2. BGP Example . . . . . . . . . . . . . . . . . . . . . . . 25 - A.2.1. Datastores . . . . . . . . . . . . . . . . . . . . . 27 - A.2.2. Adding a Peer . . . . . . . . . . . . . . . . . . . . 27 - A.2.3. Removing a Peer . . . . . . . . . . . . . . . . . . . 28 - A.3. Interface Example . . . . . . . . . . . . . . . . . . . . 29 - A.3.1. Pre-provisioned Interfaces . . . . . . . . . . . . . 29 - A.3.2. System-provided Interface . . . . . . . . . . . . . . 30 - Appendix B. Ephemeral Dynamic Datastore Example . . . . . . . . 31 - Appendix C. Implications on Data Models . . . . . . . . . . . . 32 - C.1. Proposed migration of existing YANG Data Models . . . . . 33 - C.2. Standardization of new YANG Data Models . . . . . . . . . 34 - Appendix D. Implications on other Documents . . . . . . . . . . 34 - D.1. Implications on YANG . . . . . . . . . . . . . . . . . . 34 - D.2. Implications on YANG Library . . . . . . . . . . . . . . 34 - D.3. Implications to YANG Guidelines . . . . . . . . . . . . . 35 - D.3.1. Nodes with different config/state value sets . . . . 35 - D.3.2. Auto-configured or Auto-negotiated Values . . . . . . 35 - D.4. Implications on NETCONF . . . . . . . . . . . . . . . . . 35 - D.4.1. Introduction . . . . . . . . . . . . . . . . . . . . 36 - D.4.2. Overview of additions to NETCONF . . . . . . . . . . 36 - D.4.3. Overview of NETCONF version 2 . . . . . . . . . . . . 37 - D.5. Implications on RESTCONF . . . . . . . . . . . . . . . . 40 - D.5.1. Introduction . . . . . . . . . . . . . . . . . . . . 40 - D.5.2. Overview of additions to RESTCONF . . . . . . . . . . 40 - D.5.3. Overview of a possible new RESTCONF version . . . . . 42 - Appendix E. Open Issues . . . . . . . . . . . . . . . . . . . . 43 - Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 44 + 4.1. The Startup Configuration Datastore () . . . . . 9 + 4.2. The Candidate Configuration Datastore () . . . 10 + 4.3. The Running Configuration Datastore () . . . . . 10 + 4.4. The Intended Configuration Datastore () . . . . 10 + 4.5. Conventional Configuration Datastores . . . . . . . . . . 10 + 4.6. Dynamic Datastores . . . . . . . . . . . . . . . . . . . 11 + 4.7. The Operational State Datastore () . . . . . 11 + 4.7.1. Missing Resources . . . . . . . . . . . . . . . . . . 12 + 4.7.2. System-controlled Resources . . . . . . . . . . . . . 12 + 4.7.3. Origin Metadata Annotation . . . . . . . . . . . . . 12 + 5. Implications on YANG . . . . . . . . . . . . . . . . . . . . 14 + 5.1. XPath Context . . . . . . . . . . . . . . . . . . . . . . 14 + 6. YANG Modules . . . . . . . . . . . . . . . . . . . . . . . . 15 + 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 20 + 7.1. Updates to the IETF XML Registry . . . . . . . . . . . . 20 + 7.2. Updates to the YANG Module Names Registry . . . . . . . . 20 + 8. Security Considerations . . . . . . . . . . . . . . . . . . . 20 + 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 21 + 10. References . . . . . . . . . . . . . . . . . . . . . . . . . 21 + 10.1. Normative References . . . . . . . . . . . . . . . . . . 21 + 10.2. Informative References . . . . . . . . . . . . . . . . . 22 + Appendix A. Guidelines for Defining Datastores . . . . . . . . . 23 + A.1. Define which YANG modules can be used in the datastore . 23 + A.2. Define which subset of YANG-modeled data applies . . . . 23 + A.3. Define how data is actualized . . . . . . . . . . . . . . 23 + A.4. Define which protocols can be used . . . . . . . . . . . 23 + A.5. Define YANG identities for the datastore . . . . . . . . 24 + Appendix B. Ephemeral Dynamic Datastore Example . . . . . . . . 24 + Appendix C. Example Data . . . . . . . . . . . . . . . . . . . . 25 + C.1. System Example . . . . . . . . . . . . . . . . . . . . . 26 + C.2. BGP Example . . . . . . . . . . . . . . . . . . . . . . . 28 + C.2.1. Datastores . . . . . . . . . . . . . . . . . . . . . 30 + C.2.2. Adding a Peer . . . . . . . . . . . . . . . . . . . . 30 + C.2.3. Removing a Peer . . . . . . . . . . . . . . . . . . . 31 + C.3. Interface Example . . . . . . . . . . . . . . . . . . . . 32 + C.3.1. Pre-provisioned Interfaces . . . . . . . . . . . . . 32 + C.3.2. System-provided Interface . . . . . . . . . . . . . . 33 + Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 34 1. Introduction This document provides an architectural framework for datastores as they are used by network management protocols such as NETCONF [RFC6241], RESTCONF [RFC8040] and the YANG [RFC7950] data modeling language. Datastores are a fundamental concept binding network management data models to network management protocols. Agreement on a common architectural model of datastores ensures that data models can be written in a network management protocol agnostic way. This architectural framework identifies a set of conceptual datastores but it does not mandate that all network management protocols expose all these conceptual datastores. This architecture is agnostic with regard to the encoding used by network management protocols. 2. Terminology - The keywords "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]. - This document defines the following terms: - o configuration data: Data that determines how a device behaves. - This data is modeled in YANG using "config true" nodes. - Configuration data can originate from different sources. - - o static configuration data: Configuration data that is eventually - persistent and used to get a device from its initial default state - into its desired operational state. - - o dynamic configuration data: Configuration data that is obtained - dynamically during the operation of a device through interaction - with other systems and not persistent. - - o system configuration data: Configuration data that is supplied by - the device itself. - - o default configuration data: Configuration data that is not - explicitly provided but for which a value defined in the data - model is used. - - o applied configuration data: Configuration data that is currently - used by a device. Applied configuration data consists of static - configuration data and dynamic configuration data. - - o state data: The additional data on a system that is not - configuration data such as read-only status information and - collected statistics. State data is transient and modified by - interactions with internal components or other systems. State - data is modeled in YANG using "config false" nodes. - o datastore: A conceptual place to store and access information. A datastore might be implemented, for example, using files, a database, flash memory locations, or combinations thereof. A datastore maps to an instantiated YANG data tree. - o configuration datastore: A datastore holding static configuration - data that is required to get a device from its initial default - state into a desired operational state. A configuration datastore - maps to an instantiated YANG data tree consisting of configuration - data nodes and interior data nodes. + o configuration: Data that determines how a device behaves. This + data is modeled in YANG using "config true" nodes. Configuration + can originate from different sources. - o running configuration datastore: A configuration datastore holding - the complete static configuration currently active on the device. - The running configuration datastore always exists. It may include - inactive configuration or template-mechanism-oriented - configuration that require further expansion. + o configuration datastore: A datastore holding configuration. - o intended configuration datastore: A configuration datastore - holding the complete configuration currently active on the device. - It does not include inactive configuration and it does include the - expansion of any template mechanisms. + o running configuration datastore: A configuration datastore holding + the current configuration of the device. It may include inactive + configuration or template-mechanism-oriented configuration that + require further expansion. This datastore is commonly referred to + as "". o candidate configuration datastore: A configuration datastore that can be manipulated without impacting the device's running configuration datastore and that can be committed to the running - configuration datastore. A candidate datastore may not be - supported by all protocols or implementations. + configuration datastore. This datastore is commonly referred to + as "". - o startup configuration datastore: The configuration datastore - holding the configuration loaded by the device into the running - configuration datastore when it boots. A startup datastore may - not be supported by all protocols or implementations. + o startup configuration datastore: A configuration datastore holding + the configuration loaded by the device into the running + configuration datastore when it boots. This datastore is commonly + referred to as "". - o dynamic datastore: A datastore holding dynamic configuration data. + o intended configuration: Configuration that is intended to be used + by the device. For example, intended configuration excludes any + inactive configuration and it would include configuration produced + through the expansion of templates. - o operational state datastore: A datastore holding the currently - active applied configuration data as well as the device's state - data. + o intended configuration datastore: A configuration datastore + holding the complete intended configuration of the device. This + datastore is commonly referred to as "". + + o conventional configuration datastore: One of the following set of + configuration datastores: , , , and + . These datastores share a common schema and protocol + operations allow copying data between these datastores. The term + "conventional" is chosen as a generic umbrella term for these + datastores. + + o conventional configuration: Configuration that is stored in any of + the conventional configuration datastores. + + o dynamic datastore: A datastore holding data obtained dynamically + during the operation of a device through interaction with other + systems, rather than through one of the conventional configuration + datastores. + + o dynamic configuration: Configuration obtained via a dynamic + datastore. + + o learned configuration: Configuration that has been learned via + protocol interactions with other systems that is not conventional + or dynamic configuration. + + o system configuration: Configuration that is supplied by the device + itself. + + o default configuration: Configuration that is not explicitly + provided but for which a value defined in the data model is used. + + o applied configuration: Configuration that is actively in use by a + device. Applied configuration originates from conventional, + dynamic, learned, system and default configuration. + + o system state: The additional data on a system that is not + configuration, such as read-only status information and collected + statistics. System state is transient and modified by + interactions with internal components or other systems. System + state is modeled in YANG using "config false" nodes. + + o operational state: The combination of applied configuration and + system state. + + o operational state datastore: A datastore holding the complete + operational state of the device. This datastore is commonly + referred to as "". o origin: A metadata annotation indicating the origin of a data item. - o remnant data: Configuration data that remains in the system for a - period of time after it has be removed from a configuration - datastore. The time period may be minimal, or may last until all - resources used by the newly-deleted configuration data (e.g., - network connections, memory allocations, file handles) have been + o remnant configuration: Configuration that remains part of the + applied configuration for a period of time after it has been + removed from the intended configuration or dynamic configuration. + The time period may be minimal, or may last until all resources + used by the newly-deleted configuration (e.g., network + connections, memory allocations, file handles) have been deallocated. The following additional terms are not datastore specific but commonly used and thus defined here as well: o client: An entity that can access YANG-defined data on a server, over some network management protocol. o server: An entity that provides access to YANG-defined data to a client, over some network management protocol. o notification: A server-initiated message indicating that a certain event has been recognized by the server. o remote procedure call: An operation that can be invoked by a client on a server. -3. Introduction +3. Background NETCONF [RFC6241] provides the following definitions: o datastore: A conceptual place to store and access information. A datastore might be implemented, for example, using files, a database, flash memory locations, or combinations thereof. o configuration datastore: The datastore holding the complete set of - configuration data that is required to get a device from its - initial default state into a desired operational state. + configuration that is required to get a device from its initial + default state into a desired operational state. YANG 1.1 [RFC7950] provides the following refinements when NETCONF is used with YANG (which is the usual case but note that NETCONF was - defined before YANG did exist): + defined before YANG existed): o datastore: When modeled with YANG, a datastore is realized as an instantiated data tree. o configuration datastore: When modeled with YANG, a configuration datastore is realized as an instantiated data tree with - configuration data. + configuration. [RFC6244] defined operational state data as follows: o Operational state data is a set of data that has been obtained by the system at runtime and influences the system's behavior similar to configuration data. In contrast to configuration data, operational state is transient and modified by interactions with internal components or other systems via specialized protocols. Section 4.3.3 of [RFC6244] discusses operational state and among other things mentions the option to consider operational state as being stored in another datastore. Section 4.4 of this document then - concludes that at the time of the writing, modeling state as a - separate data tree is the recommended approach. + concludes that at the time of the writing, modeling state as distinct + leafs and distinct branches is the recommended approach. Implementation experience and requests from operators [I-D.ietf-netmod-opstate-reqs], [I-D.openconfig-netmod-opstate] indicate that the datastore model initially designed for NETCONF and refined by YANG needs to be extended. In particular, the notion of intended configuration and applied configuration has developed. - Furthermore, separating operational state data from configuration - data in a separate branch in the data model has been found - operationally complicated, and typically impacts the readability of - module definitions due to overuse of groupings. The relationship - between the branches is not machine readable and filter expressions - operating on configuration data and on related operational state data - are different. + Furthermore, separating operational state from configuration in a + separate branch in the data model has been found operationally + complicated, and typically impacts the readability of module + definitions due to overuse of groupings. The relationship between + the branches is not machine readable and filter expressions operating + on configuration and on related operational state are different. 3.1. Original Model of Datastores The following drawing shows the original model of datastores as it is currently used by NETCONF [RFC6241]: +-------------+ +-----------+ | | | | | (ct, rw) |<---+ +--->| (ct, rw) | +-------------+ | | +-----------+ @@ -298,53 +294,52 @@ v operational state <--- control plane (cf, ro) ct = config true; cf = config false rw = read-write; ro = read-only boxes denote datastores Note that this diagram simplifies the model: read-only (ro) and read- write (rw) is to be understood at a conceptual level. In NETCONF, - for example, support for the and datastores is - optional and the datastore does not have to be writable. - Furthermore, the datastore can only be modified by copying - to in the standardized NETCONF datastore editing - model. The RESTCONF protocol does not expose these differences and - instead provides only a writable unified datastore, which hides - whether edits are done through a datastore or by directly - modifying the datastore or via some other implementation - specific mechanism. RESTCONF also hides how configuration is made - persistent. Note that implementations may also have additional - datastores that can propagate changes to the datastore. + for example, support for and is optional and + does not have to be writable. Furthermore, can + only be modified by copying to in the + standardized NETCONF datastore editing model. The RESTCONF protocol + does not expose these differences and instead provides only a + writable unified datastore, which hides whether edits are done + through or by directly modifying or via some + other implementation specific mechanism. RESTCONF also hides how + configuration is made persistent. Note that implementations may also + have additional datastores that can propagate changes to . NETCONF explicitly mentions so called named datastores. Some observations: o Operational state has not been defined as a datastore although there were proposals in the past to introduce an operational state datastore. o The NETCONF operation returns the content of the configuration datastore together with the operational state. It - is therefore necessary that config false data is in a different - branch than the config true data if the operational state data can - have a different lifetime compared to configuration data or if - configuration data is not immediately or successfully applied. + is therefore necessary that "config false" data is in a different + branch than the "config true" data if the operational state can + have a different lifetime compared to configuration or if + configuration is not immediately or successfully applied. o Several implementations have proprietary mechanisms that allow - clients to store inactive data in the datastore; this - inactive data is only exposed to clients that indicate that they - support the concept of inactive data; clients not indicating - support for inactive data receive the content of the - datastore with the inactive data removed. Inactive data is - conceptually removed before validation. + clients to store inactive data in ; this inactive data is + only exposed to clients that indicate that they support the + concept of inactive data; clients not indicating support for + inactive data receive the content of with the inactive + data removed. Inactive data is conceptually removed before + validation. o Some implementations have proprietary mechanisms that allow clients to define configuration templates in . These templates are expanded automatically by the system, and the resulting configuration is applied internally. o Some operators have reported that it is essential for them to be able to retrieve the configuration that has actually been successfully applied, which may be a subset or a superset of the configuration. @@ -358,256 +353,297 @@ +-------------+ +-----------+ | | | | | (ct, rw) |<---+ +--->| (ct, rw) | +-------------+ | | +-----------+ | | | | | +-----------+ | +-------->| |<--------+ | (ct, rw) | +-----------+ | + | // configuration transformations, | // e.g., removal of "inactive" | // nodes, expansion of templates v +------------+ | | // subject to validation | (ct, ro) | +------------+ + | // changes applied, subject to + | // local factors, e.g., missing + | // resources, delays | - | // e.g., missing resources, delays - | - | +------ auto-discovery - | +------ dynamic configuration protocols - | +------ control-plane protocols - | +------ dynamic datastores - | | - v v + | +-------- learned configuration + dynamic | +-------- system configuration + datastores -----+ | +-------- default configuration + | | | + v v v +---------------+ - | | + | | <-- system state | (ct + cf, ro) | +---------------+ ct = config true; cf = config false rw = read-write; ro = read-only - boxes denote datastores + boxes denote named datastores -4.1. The Datastore +4.1. The Startup Configuration Datastore () - The datastore is a read-only datastore that consists of - config true nodes. It is tightly coupled to . When data is - written to , the data that is to be validated is also - conceptually written to . Validation is performed on the - contents of . + The startup configuration datastore () is an optional + configuration datastore holding the configuration loaded by the + device when it boots. is only present on devices that + separate the startup configuration from the running configuration + datastore. - On a traditional NETCONF implementation, and are - always the same. + The startup configuration datastore may not be supported by all + protocols or implementations. + +4.2. The Candidate Configuration Datastore () + + The candidate configuration datastore () is an optional + configuration datastore that can be manipulated without impacting the + device's current configuration and that can be committed to + . + + The candidate configuration datastore may not be supported by all + protocols or implementations. + +4.3. The Running Configuration Datastore () + + The running configuration datastore () holds the complete + current configuration on the device. It may include inactive + configuration or template-mechanism-oriented configuration that + require further expansion. + +4.4. The Intended Configuration Datastore () + + The intended configuration datastore () is a read-only + configuration datastore. It is tightly coupled to . When + data is written to , the data that is to be validated is + also conceptually written to . Validation is performed on + the contents of . + + For simple implementations, and are identical. Currently there are no standard mechanisms defined that affect so that it would have different contents than , but this architecture allows for such mechanisms to be defined. One example of such a mechanism is support for marking nodes as inactive in . Inactive nodes are not copied to , and are thus not taken into account when validating the configuration. Another example is support for templates. Templates are expanded when copied into , and the expanded result is validated. -4.2. Dynamic Datastores +4.5. Conventional Configuration Datastores - The model recognizes the need for dynamic datastores that are by - definition not part of the persistent configuration of a device. In + The conventional configuration datastores are a set of configuration + datastores that share a common schema, allowing data to be copied + between them. The term is meant as a generic umbrella description of + these datastores. The set of datastores include: + + o + o + + o + + o + + Other conventional configuration datastores may be defined in future + documents. + + The flow of data between these datastores is depicted in section + Section 4. + + The specific protocols may define explicit operations to copy between + these datastores, e.g., NETCONF's operation. + +4.6. Dynamic Datastores + + The model recognizes the need for dynamic datastores that are, by + definition, not part of the persistent configuration of a device. In some contexts, these have been termed ephemeral datastores since the information is ephemeral, i.e., lost upon reboot. The dynamic - datastores interact with the rest of the system through the - datastore. - - Note that the ephemeral datastore discussed in I2RS documents maps to - a dynamic datastore in the datastore model described here. + datastores interact with the rest of the system through + . -4.3. The Datastore +4.7. The Operational State Datastore () - The datastore is a read-only datastore that consists of - config true and config false nodes. In the original NETCONF model - the operational state only had config false nodes. The reason for - incorporating config true nodes here is to be able to expose all - operational settings without having to replicate definitions in the - data models. + The operational state datastore () is a read-only + datastore that consists of all "config true" and "config false" nodes + defined in the schema. In the original NETCONF model the operational + state only had "config false" nodes. The reason for incorporating + "config true" nodes here is to be able to expose all operational + settings without having to replicate definitions in the data models. - The datastore contains all configuration data actually - used by the system, including all applied configuration, system- - provided configuration and values defined by any supported data - models. In addition, the datastore also contains state - data. + contains system state and all configuration actually + used by the system. This includes all applied configuration from + , system-provided configuration, and default values defined + by any supported data models. In addition, also + contains applied data from dynamic datastores. - Changes to configuration data may take time to percolate through to - the datastore. During this period, the - datastore will return data nodes for both the previous and current - configuration, as closely as possible tracking the current operation - of the device. These "remnants" of the previous configuration - persist while the system has released resources used by the newly- - deleted configuration data (e.g., network connections, memory - allocations, file handles). + Changes to configuration may take time to percolate through to + . During this period, may contain nodes + for both the previous and current configuration, as closely as + possible tracking the current operation of the device. Such remnant + configuration from the previous configuration persists until the + system has released resources used by the newly-deleted configuration + (e.g., network connections, memory allocations, file handles). - As a result of these remnants, the semantic constraints defined in - the data model cannot be relied upon for the datastore, - since the system may have remnants whose constraints were valid with - the previous configuration and that are not valid with the current - configuration. Since constraints on "config false" nodes may refer - to "config true" nodes, remnants may force the violation of those - constraints. The constraints that may not hold include "when", - "must", "min-elements", and "max-elements". Note that syntactic - constraints cannot be violated, including hierarchical organization, - identifiers, and type-based constraints. + As a result of remnant configuration, the semantic constraints + defined in the data model cannot be relied upon for , + since the system may have remnant configuration whose constraints + were valid with the previous configuration and that are not valid + with the current configuration. Since constraints on "config false" + nodes may refer to "config true" nodes, remnant configuration may + force the violation of those constraints. The constraints that may + not hold include "when", "must", "min-elements", and "max-elements". + Note that syntactic constraints cannot be violated, including + hierarchical organization, identifiers, and type-based constraints. -4.3.1. Missing Resources +4.7.1. Missing Resources - The configuration can refer to resources that are not + Configuration in can refer to resources that are not available or otherwise not physically present. In these situations, these parts of the configuration are not applied. The data appears in but does not appear in . A typical example is an interface configuration that refers to an interface that is not currently present. In such a situation, the interface configuration remains in but the interface configuration will not appear in . Note that configuration validity cannot depend on the current state of such resources, since that would imply the removing a resource might render the configuration invalid. This is unacceptable, especially given that rebooting such a device would fail to boot due to an invalid configuration. Instead we allow configuration for missing resources to exist in and , but it will not appear in . -4.3.2. System-controlled Resources +4.7.2. System-controlled Resources Sometimes resources are controlled by the device and the corresponding system controlled data appear in (and disappear from) dynamically. If a system controlled resource has matching configuration in when it appears, the system will try to apply the configuration, which causes the configuration to appear in eventually (if application of the configuration was successful). -4.3.3. Origin Metadata Annotation - - As data flows into the datastore, it is conceptually - marked with a metadata annotation ([RFC7952]) that indicates its - origin. The "origin" metadata annotation is defined in Section 6. - The values are YANG identities. The following identities are - defined: - - +-- origin - +-- static - +-- dynamic - +-- default - +-- system +4.7.3. Origin Metadata Annotation - These identities can be further refined, e.g., there might be an - identity "dhcp" derived from "dynamic". + As data flows into , it is conceptually marked with a + metadata annotation ([RFC7952]) that indicates its origin. The + origin applies to all data nodes except non-presence containers. The + "origin" metadata annotation is defined in Section 6. The values are + YANG identities. The following identities are defined: - The "static" origin represents data provided by the - datastore. The "dynamic" origin represents data provided by a - dynamic datastore. The "default" origin represents data values - specified in the data model, using either simple values in the - "default" statement or any values described in the "description" - statement. Finally, the "system" origin represents data learned from - the normal operational of the system, including control-plane - protocols. + o origin: abstract base identity from which the other origin + identities are derived. -5. Guidelines for Defining Dynamic Datastores + o intended: represents data provided by . - The definition of a dynamic datastore SHOULD be provided in a - document (e.g., an RFC) purposed to the definition of the dynamic - datastore. When it makes sense, more than one dynamic datastore MAY - be defined in the same document (e.g., when the datastores are - logically connected). Each dynamic datastore's definition SHOULD - address the points specified in the sections below. + o dynamic: represents data provided by a dynamic datastore. -5.1. Define a name for the dynamic datastore + o system: represents data provided by the system itself, including + both system configuration and system state. Examples of system + configuration include applied configuration for an always existing + loopback interface, or interface configuration that is auto- + created due to the hardware currently present in the device. - Each dynamic datastores MUST have a name using the character set - described by Section 6.2 of [RFC7950]. The name SHOULD be consistent - in style and length to other datastore names described in this - document. + o learned: represents configuration that has been learned via + protocol interactions with other systems, including protocols such + as link-layer negotiations, routing protocols, DHCP, etc. - The datastore's name does not need to be globally unique, as it will - be uniquely qualified by the namespace of the module in which it is - defined (Section 5.6). This means that names such as "running" and - "operational" are valid datastore names. However, it is usually - desirable to avoid using the same name for multiple different - datastores. + o default: represents data using a default value specified in the + data model, using either values in the "default" statement or any + values described in the "description" statement. The default + origin is only used when the data has not been provided by any + other source. -5.2. Define which YANG modules can be used in the datastore + o unknown: represents data for which the system cannot identify the + origin. - Not all YANG modules may be used in all datastores. Some datastores - may constrain which data models can be used in them. If it is - desirable that a subset of all modules can be targeted to the dynamic - datastore, then the documentation defining the dynamic datastore MUST - use the mechanisms described in Appendix D.2 to provide the necessary - hooks for module-designers to indicate that their module is to be - accessible in the dynamic datastore. + These identities can be further refined, e.g., there could be + separate identities for particular types or instances of dynamic + datastore derived from "dynamic". -5.3. Define which subset of YANG-modeled data applies + In all cases, the device should report the origin that most + accurately reflects the source of the data that is actively being + used by the system. - By default, the data in a dynamic datastore is modeled by all YANG - statements in the available YANG modules. However, it is possible to - specify criteria YANG statements must satisfy in order to be present - in a dynamic datastore. For instance, maybe only config true nodes - are present, or config false nodes that also have a specific YANG - extension (e.g., i2rs:ephemeral true) are present in the dynamic - datastore. + In cases where it could be ambiguous as to which origin should be + used, i.e. where the same data node value has originated from + multiple sources, then the description statement in the YANG module + should be used as guidance for choosing the appropriate origin. For + example: -5.4. Define how dynamic data is actualized + If for a particular configuration node, the associated YANG + description statement indicates that a protocol negotiated value + overrides any configured value, then the origin would be reported as + "learned", even when a learned value is the same as the configured + value. - The diagram in Section 4 depicts dynamic datastores feeding into the - datastore. How this interaction occurs must be defined - by the dynamic datastore. In some cases, it may occur implicitly, as - soon as the data is put into the dynamic datastore while, in other - cases, an explicit action (e.g., an RPC) may be required to trigger - the application of the dynamic datastore's data. + Conversely, if for a particular configuration node, the associated + YANG description statement indicates that a protocol negotiated value + does not override an explicitly configured value, then the origin + would be reported as "intended" even when a learned value is the same + as the configured value. -5.5. Define which protocols can be used + In the case that a device cannot provide an accurate origin for a + particular data node then it should use the origin "unknown". - By default, it is assumed that both the NETCONF and RESTCONF - protocols can be used to interact with a dynamic datastore. However, - it may be that only a specific protocol can be used (e.g., Forces) or - that a subset of all protocol operations or capabilities are - available (e.g., no locking, no xpath-based filtering, etc.). +5. Implications on YANG -5.6. Define a module for the dynamic datastore +5.1. XPath Context - Each dynamic datastore MUST be defined by a YANG module. This module - is used by servers to indicate (e.g., via YANG Library) their support - for the dynamic datastore. + If a server implements the architecture defined in this document, the + accessible trees for some XPath contexts are refined as follows: - The YANG module MUST import the "ietf-datastores" and "ietf-origin" - modules, defined in this document. This is necessary in order to - access the base identities they define. + o If the XPath expression is defined in a substatement to a data + node that represents system state, the accessible tree is all + operational state in the server. The root node has all top-level + data nodes in all modules as children. - The YANG module MUST define an identity that uses the "ds:datastore" - identity as its base. This identity is necessary so that the - datastore can be referenced in protocol operations (e.g., - ). + o If the XPath expression is defined in a substatement to a + "notification" statement, the accessible tree is the notification + instance and all operational state in the server. If the + notification is defined on the top level in a module, then the + root node has the node representing the notification being defined + and all top-level data nodes in all modules as children. + Otherwise, the root node has all top-level data nodes in all + modules as children. - The YANG module MUST define an identity that uses the "or:dynamic" - identity as its base. This identity is necessary so that data - originating from the datastore can be identified as such via the - "origin" metadata attribute defined in Section 6. + o If the XPath expression is defined in a substatement to an "input" + statement in an "rpc" or "action" statement, the accessible tree + is the RPC or action operation instance and all operational state + in the server. The root node has top-level data nodes in all + modules as children. Additionally, for an RPC, the root node also + has the node representing the RPC operation being defined as a + child. The node representing the operation being defined has the + operation's input parameters as children. - An example of these guidelines in use is provided in Appendix B. + o If the XPath expression is defined in a substatement to an + "output" statement in an "rpc" or "action" statement, the + accessible tree is the RPC or action operation instance and all + operational state in the server. The root node has top-level data + nodes in all modules as children. Additionally, for an RPC, the + root node also has the node representing the RPC operation being + defined as a child. The node representing the operation being + defined has the operation's output parameters as children. 6. YANG Modules - file "ietf-datastores@2017-03-13.yang" + file "ietf-datastores@2017-04-26.yang" module ietf-datastores { yang-version 1.1; namespace "urn:ietf:params:xml:ns:yang:ietf-datastores"; prefix ds; organization "IETF NETMOD (NETCONF Data Modeling Language) Working Group"; contact @@ -642,81 +678,85 @@ without modification, is permitted pursuant to, and subject to the license terms contained in, the Simplified BSD License set forth in Section 4.c of the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info). This version of this YANG module is part of RFC XXXX (http://www.rfc-editor.org/info/rfcxxxx); see the RFC itself for full legal notices."; - revision 2017-03-13 { + revision 2017-04-26 { description "Initial revision."; reference "RFC XXXX: Network Management Datastore Architecture"; } /* * Identities */ identity datastore { description "Abstract base identity for datastore identities."; } - identity static { + identity conventional { + base datastore; description - "Abstract base identity for static configuration datastores."; + "Abstract base identity for conventional configuration + datastores."; } identity dynamic { + base datastore; description - "Abstract base identity for dynamic configuration datastores."; + "Abstract base identity for dynamic datastores."; } identity running { - base static; + base conventional; description - "The 'running' datastore."; + "The running configuration datastore."; } identity candidate { - base static; + base conventional; description - "The 'candidate' datastore."; + "The candidate configuration datastore."; } identity startup { - base static; + base conventional; description - "The 'startup' datastore."; + "The startup configuration datastore."; + } identity intended { - base static; + base conventional; description - "The 'intended' datastore."; + "The intended configuration datastore."; } identity operational { base datastore; description - "The 'operational' state datastore."; + "The operational state datastore."; } } - file "ietf-datastores@2017-03-13.yang" + file "ietf-origin@2017-04-26.yang" module ietf-origin { yang-version 1.1; namespace "urn:ietf:params:xml:ns:yang:ietf-origin"; prefix or; import ietf-yang-metadata { prefix md; } @@ -738,90 +778,114 @@ Author: Kent Watsen Author: Rob Wilton "; description "This YANG module defines an 'origin' metadata annotation, and a - set of identities for the origin value. The 'origin' metadata - annotation is used to mark data in the 'operational' - datastore with information on where the data originated. + set of identities for the origin value. Copyright (c) 2017 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 Simplified BSD License set forth in Section 4.c of the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info). This version of this YANG module is part of RFC XXXX (http://www.rfc-editor.org/info/rfcxxxx); see the RFC itself for full legal notices."; - revision 2017-03-13 { + revision 2017-04-26 { description "Initial revision."; reference "RFC XXXX: Network Management Datastore Architecture"; } /* * Identities */ identity origin { description "Abstract base identity for the origin annotation."; } - identity static { + identity intended { base origin; description - "Denotes data from static configuration (e.g., )."; + "Denotes data from the intended configuration datastore"; } + identity dynamic { base origin; description - "Denotes data from dynamic configuration protocols - or dynamic datastores (e.g., DHCP)."; + "Denotes data from a dynamic datastore."; } - identity system { base origin; description - "Denotes data created by the system independently of what - has been configured."; + "Denotes data originated by the system itself, including + both system configuration and system state. + + Examples of system configuration include applied configuration + for an always existing loopback interface, or interface + configuration that is auto-created due to the hardware + currently present in the device."; + } + + identity learned { + base origin; + description + "Denotes configuration learned from protocol interactions with + other devices, instead of via the intended configuration + datastore or any dynamic datastore. + + Examples of protocols that provide learned configuration + include link-layer negotiations, routing protocols, and + DHCP."; } identity default { base origin; description - "Denotes data that does not have an explicitly configured - value, but has a default value in use. Covers both simple - defaults and defaults defined via an explanation in a - description statement."; + "Denotes data that does not have an configured or learned + value, but has a default value in use. Covers both values + defined in a 'default' statement, and values defined via an + explanation in a 'description' statement."; + } + + identity unknown { + base origin; + description + "Denotes data for which the system cannot identify the + origin."; } /* * Metadata annotations */ md:annotation origin { type identityref { base origin; } + description + "The 'origin' annotation can be present on any node in a + datastore. It specifies from where the node originated."; } } 7. IANA Considerations 7.1. Updates to the IETF XML Registry @@ -848,23 +912,23 @@ prefix: ds reference: RFC XXXX name: ietf-origin namespace: urn:ietf:params:xml:ns:yang:ietf-origin prefix: or reference: RFC XXXX 8. Security Considerations - This document discusses a conceptual model of datastores for network - management using NETCONF/RESTCONF and YANG. It has no security - impact on the Internet. + This document discusses an architectural model of datastores for + network management using NETCONF/RESTCONF and YANG. It has no + security impact on the Internet. 9. Acknowledgments This document grew out of many discussions that took place since 2010. Several Internet-Drafts ([I-D.bjorklund-netmod-operational], [I-D.wilton-netmod-opstate-yang], [I-D.ietf-netmod-opstate-reqs], [I-D.kwatsen-netmod-opstate], [I-D.openconfig-netmod-opstate]) and [RFC6244] touched on some of the problems of the original datastore model. The following people were authors to these Internet-Drafts or otherwise actively involved in the discussions that led to this @@ -888,34 +953,25 @@ o Rob Shakir, Google, Juergen Schoenwaelder was partly funded by Flamingo, a Network of Excellence project (ICT-318488) supported by the European Commission under its Seventh Framework Programme. 10. References 10.1. Normative References - [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate - Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/ - RFC2119, March 1997, - . - [RFC6241] Enns, R., Ed., Bjorklund, M., Ed., Schoenwaelder, J., Ed., and A. Bierman, Ed., "Network Configuration Protocol (NETCONF)", RFC 6241, DOI 10.17487/RFC6241, June 2011, . - [RFC7895] Bierman, A., Bjorklund, M., and K. Watsen, "YANG Module - Library", RFC 7895, DOI 10.17487/RFC7895, June 2016, - . - [RFC7950] Bjorklund, M., Ed., "The YANG 1.1 Data Modeling Language", RFC 7950, DOI 10.17487/RFC7950, August 2016, . [RFC7952] Lhotka, L., "Defining and Using Metadata with YANG", RFC 7952, DOI 10.17487/RFC7952, August 2016, . [RFC8040] Bierman, A., Bjorklund, M., and K. Watsen, "RESTCONF Protocol", RFC 8040, DOI 10.17487/RFC8040, January 2017, @@ -926,25 +982,20 @@ [I-D.bjorklund-netmod-operational] Bjorklund, M. and L. Lhotka, "Operational Data in NETCONF and YANG", draft-bjorklund-netmod-operational-00 (work in progress), October 2012. [I-D.ietf-netmod-opstate-reqs] Watsen, K. and T. Nadeau, "Terminology and Requirements for Enhanced Handling of Operational State", draft-ietf- netmod-opstate-reqs-04 (work in progress), January 2016. - [I-D.ietf-netmod-rfc6087bis] - Bierman, A., "Guidelines for Authors and Reviewers of YANG - Data Model Documents", draft-ietf-netmod-rfc6087bis-12 - (work in progress), March 2017. - [I-D.kwatsen-netmod-opstate] Watsen, K., Bierman, A., Bjorklund, M., and J. Schoenwaelder, "Operational State Enhancements for YANG, NETCONF, and RESTCONF", draft-kwatsen-netmod-opstate-02 (work in progress), February 2016. [I-D.openconfig-netmod-opstate] Shakir, R., Shaikh, A., and M. Hines, "Consistent Modeling of Operational State Data in YANG", draft-openconfig- netmod-opstate-01 (work in progress), July 2015. @@ -956,36 +1007,146 @@ [RFC3688] Mealling, M., "The IETF XML Registry", BCP 81, RFC 3688, DOI 10.17487/RFC3688, January 2004, . [RFC6020] Bjorklund, M., Ed., "YANG - A Data Modeling Language for the Network Configuration Protocol (NETCONF)", RFC 6020, DOI 10.17487/RFC6020, October 2010, . - [RFC6243] Bierman, A. and B. Lengyel, "With-defaults Capability for - NETCONF", RFC 6243, DOI 10.17487/RFC6243, June 2011, - . - [RFC6244] Shafer, P., "An Architecture for Network Management Using NETCONF and YANG", RFC 6244, DOI 10.17487/RFC6244, June 2011, . -Appendix A. Example Data +Appendix A. Guidelines for Defining Datastores + + The definition of a new datastore in this architecture should be + provided in a document (e.g., an RFC) purposed to the definition of + the datastore. When it makes sense, more than one datastore may be + defined in the same document (e.g., when the datastores are logically + connected). Each datastore's definition should address the points + specified in the sections below. + +A.1. Define which YANG modules can be used in the datastore + + Not all YANG modules may be used in all datastores. Some datastores + may constrain which data models can be used in them. If it is + desirable that a subset of all modules can be targeted to the + datastore, then the documentation defining the datastore must + indicate this. + +A.2. Define which subset of YANG-modeled data applies + + By default, the data in a datastore is modeled by all YANG statements + in the available YANG modules. However, it is possible to specify + criteria that YANG statements must satisfy in order to be present in + a datastore. For instance, maybe only "config true" nodes are + present, or "config false nodes" that also have a specific YANG + extension (e.g., "i2rs:ephemeral true") are present in the datastore. + +A.3. Define how data is actualized + + The new datastore must specify how it interacts with other + datastores. For example, the diagram in Section 4 depicts dynamic + datastores feeding into . How this interaction occurs + must be defined by any dynamic datastore. In some cases, it may + occur implicitly, as soon as the data is put into the dynamic + datastore while, in other cases, an explicit action (e.g., an RPC) + may be required to trigger the application of the datastore's data. + +A.4. Define which protocols can be used + + By default, it is assumed that both the NETCONF and RESTCONF + protocols can be used to interact with a datastore. However, it may + be that only a specific protocol can be used (e.g., ForCES) or that a + subset of all protocol operations or capabilities are available + (e.g., no locking or no XPath-based filtering). + +A.5. Define YANG identities for the datastore + + The datastore must be defined with a YANG identity that uses the + "ds:datastore" identity or one of its derived identities as its base. + This identity is necessary so that the datastore can be referenced in + protocol operations (e.g., ). + + The datastore may also be defined with an identity that uses the + "or:origin" identity or one its derived identities as its base. This + identity is needed if the datastore interacts with so + that data originating from the datastore can be identified as such + via the "origin" metadata attribute defined in Section 6. + + An example of these guidelines in use is provided in Appendix B. + +Appendix B. Ephemeral Dynamic Datastore Example + + The section defines documentation for an example dynamic datastore + using the guidelines provided in Appendix A. While this example is + very terse, it is expected to be that a standalone RFC would be + needed when fully expanded. + + This example defines a dynamic datastore called "ephemeral", which is + loosely modeled after the work done in the I2RS working group. + + 1. Name : ephemeral + 2. YANG modules : all (default) + 3. YANG statements : config false + ephemeral true + 4. How applied : automatic + 5. Protocols : NC/RC (default) + 6. YANG Module : (see below) + + module example-ds-ephemeral { + yang-version 1.1; + namespace "urn:example:ds-ephemeral"; + prefix eph; + + import ietf-datastores { + prefix ds; + } + import ietf-origin { + prefix or; + } + + // add datastore identity + identity ds-ephemeral { + base ds:datastore; + description + "The 'ephemeral' datastore."; + } + + // add origin identity + identity or-ephemeral { + base or:dynamic; + description + "Denotes data from the ephemeral dynamic datastore."; + } + + // define ephemeral extension + extension ephemeral { + argument "value"; + description + "This extension is mixed into config false YANG nodes to + indicate that they are writable nodes in the 'ephemeral' + datastore. This statement takes a single argument + representing a boolean having the values 'true' and + 'false'. The default value is 'false'."; + } + } + +Appendix C. Example Data The use of datastores is complex, and many of the subtle effects are more easily presented using examples. This section presents a series of example data models with some sample contents of the various datastores. -A.1. System Example +C.1. System Example In this example, the following fictional module is used: module example-system { yang-version 1.1; namespace urn:example:system; prefix sys; import ietf-inet-types { prefix inet; @@ -1072,21 +1233,21 @@ to a default value, a loopback interface is automatically added by the system, and the result of the "speed" auto-negotiation. All of this is reflected in : bar - + eth0 true 1000 100
2001:db8::10 32
@@ -1099,21 +1260,21 @@ lo0
::1 128
-A.2. BGP Example +C.2. BGP Example Consider the following piece of a ersatz BGP module: container bgp { leaf local-as { type uint32; } leaf peer-as { type uint32; } @@ -1153,122 +1314,119 @@ In this example model, both bgp/peer/local-as and bgp/peer/peer-as have complex hierarchical values, allowing the user to specify default values for all peers in a single location. The model also follows the pattern of fully integrating state ("config false") nodes with configuration ("config true") nodes. There is not separate "bgp-state" hierarchy, with the accompanying repetition of containment and naming nodes. This makes the model simpler and more readable. -A.2.1. Datastores +C.2.1. Datastores - Each datastore represents differing views of these data nodes. The - datastore will hold the configuration data provided by the - user, for example a single BGP peer. The datastore will - conceptually hold the data as validated, after the removal of data - not intended for validation and after any local template mechanisms - are performed. The datastore will show data from - as well as any "config false" nodes. + Each datastore represents differing views of these nodes. + will hold the configuration provided by the user, for example a + single BGP peer. will conceptually hold the data as + validated, after the removal of data not intended for validation and + after any local template mechanisms are performed. + will show data from as well as any "config false" nodes. -A.2.2. Adding a Peer +C.2.2. Adding a Peer If the user configures a single BGP peer, then that peer will be - visible in both the and datastores. It may also - appear in the datastore, if the server supports the - "candidate" feature. Retrieving the peer will return only the user- - specified values. + visible in both and . It may also appear in + , if the server supports the "candidate" feature. + Retrieving the peer will return only the user-specified values. No time delay should exist between the appearance of the peer in and . In this scenario, we've added the following to : 64642 65000 10.1.2.3 -A.2.2.1. +C.2.2.1. - The datastore will contain the fully expanded peer - data, including "config false" nodes. In our example, this means the - "state" node will appear. + will contain the fully expanded peer data, including + "config false" nodes. In our example, this means the "state" node + will appear. - In addition, the datastore will contain the "currently - in use" values for all nodes. This means that local-as and peer-as - will be populated even if they are not given values in . - The value of bgp/local-as will be used if bgp/peer/local-as is not - provided; bgp/peer-as and bgp/peer/peer-as will have the same - relationship. In the operational view, this means that every peer - will have values for their local-as and peer-as, even if those values - are not explicitly configured but are provided by bgp/local-as and - bgp/peer-as. + In addition, will contain the "currently in use" values + for all nodes. This means that local-as and peer-as will be + populated even if they are not given values in . The value + of bgp/local-as will be used if bgp/peer/local-as is not provided; + bgp/peer-as and bgp/peer/peer-as will have the same relationship. In + the operational view, this means that every peer will have values for + their local-as and peer-as, even if those values are not explicitly + configured but are provided by bgp/local-as and bgp/peer-as. Each BGP peer has a TCP connection associated with it, using the - values of local-port and remote-port from the intended datastore. If - those values are not supplied, the system will select values. When - the connection is established, the datastore will - contain the current values for the local-port and remote-port nodes - regardless of the origin. If the system has chosen the values, the - "origin" attribute will be set to "operational". Before the - connection is established, one or both of the nodes may not appear, - since the system may not yet have their values. + values of local-port and remote-port from . If those + values are not supplied, the system will select values. When the + connection is established, will contain the current + values for the local-port and remote-port nodes regardless of the + origin. If the system has chosen the values, the "origin" attribute + will be set to "operational". Before the connection is established, + one or both of the nodes may not appear, since the system may not yet + have their values. - - 64642 - 65000 - - 10.1.2.3 + + 64642 + 65000 + + 10.1.2.3 64642 65000 60794 179 -A.2.3. Removing a Peer +C.2.3. Removing a Peer - Changes to configuration data may take time to percolate through the + Changes to configuration may take time to percolate through the various software components involved. During this period, it is imperative to continue to give an accurate view of the working of the - device. The datastore will return data nodes for both - the previous and current configuration, as closely as possible - tracking the current operation of the device. + device. will contain nodes for both the previous and + current configuration, as closely as possible tracking the current + operation of the device. Consider the scenario where a client removes a BGP peer. When a peer is removed, the operational state will continue to reflect the existence of that peer until the peer's resources are released, including closing the peer's connection. During this period, the - current data values will continue to be visible in the - datastore, with the "origin" attribute set to indicate the origin of - the original data. + current data values will continue to be visible in , + with the "origin" attribute set to indicate the origin of the + original data. - - 64642 - 65000 - - 10.1.2.3 + + 64642 + 65000 + + 10.1.2.3 64642 65000 - 60794 - 179 + 60794 + 179 Once resources are released and the connection is closed, the peer's - data is removed from the datastore. + data is removed from . -A.3. Interface Example +C.3. Interface Example In this section, we'll use this simple interface data model: container interfaces { list interface { key name; leaf name { type string; } leaf description { @@ -1276,685 +1434,89 @@ } leaf mtu { type uint; } leaf ipv4-address { type inet:ipv4-address; } } } -A.3.1. Pre-provisioned Interfaces +C.3.1. Pre-provisioned Interfaces One common issue in networking devices is the support of Field Replaceable Units (FRUs) that can be inserted and removed from the device without requiring a reboot or interfering with normal operation. These FRUs are typically interface cards, and the devices support pre-provisioning of these interfaces. If a client creates an interface "et-0/0/0" but the interface does - not physically exist at this point, then the datastore - might contain the following: + not physically exist at this point, then might contain the + following: et-0/0/0 Test interface - Since the interface does not exist, this data does not appear in the - datastore. + Since the interface does not exist, this data does not appear in + . When a FRU containing this interface is inserted, the system will detect it and process the associated configuration. The will contain the data from , as well as the "config false" nodes, such as the current value of the interface's MTU. - - - et-0/0/0 - Test interface + + + et-0/0/0 + Test interface 1500 - If the FRU is removed, the interface data is removed from the - datastore. + If the FRU is removed, the interface data is removed from + . -A.3.2. System-provided Interface +C.3.2. System-provided Interface Imagine if the system provides a loopback interface (named "lo0") with a default ipv4-address of "127.0.0.1". The system will only - provide configuration for this interface if the is no data for it in - . + provide configuration for this interface if there is no data for it + in . When no configuration for "lo0" appears in , then will show the system-provided data: - + lo0 127.0.0.1 When configuration for "lo0" does appear in , then will show that data with the origin set to "intended". If the "ipv4-address" is not provided, then the system-provided value will appear as follows: - - - lo0 - loopback + + + lo0 + loopback 127.0.0.1 -Appendix B. Ephemeral Dynamic Datastore Example - - The section defines documentation for an example dynamic datastore - using the guidelines provided in Section 5. While this example is - very terse, it is expected to be that a standalone RFC would be - needed when fully expanded. - - This example defines a dynamic datastore called "ephemeral", which is - loosely modeled after the work done in the I2RS working group. - - 1. Name : ephemeral - 2. YANG modules : all (default) - 3. YANG statements : config false + ephemeral true - 4. How applied : automatic - 5. Protocols : NC/RC (default) - 6. YANG Module : (see below) - - module example-ds-ephemeral { - yang-version 1.1; - namespace "urn:example:ds-ephemeral"; - prefix eph; - - import ietf-datastores { - prefix ds; - } - import ietf-origin { - prefix or; - } - - // add datastore identity - identity ds-ephemeral { - base ds:datastore; - description - "The 'ephemeral' datastore."; - } - - // add origin identity - identity or-ephemeral { - base or:dynamic; - description - "Denotes data from the ephemeral dynamic datastore."; - } - - // define ephemeral extension - extension ephemeral { - argument "value"; - description - "This extension is mixed into config false YANG nodes to - indicate that they are writable nodes in the 'ephemeral' - datastore. This statement takes a single argument - representing a boolean having the values 'true' and 'false'. - The default value is 'false'."; - } - } - -Appendix C. Implications on Data Models - - Since the NETCONF operation returns the content of the - configuration datastore and the operational state together - in one tree, data models were often forced to branch at the top-level - into a config true branch and a structurally similar config false - branch that replicated some of the config true nodes and added state - nodes. With the datastore model described here this is not needed - anymore since the different datastores handle the different lifetimes - of data objects. Introducing this model together with the - deprecation of the operation makes it possible to write - simpler models. - -C.1. Proposed migration of existing YANG Data Models - - For standards based YANG modules that have already been published, - that are using split config and state trees, it is planned that these - modules are updated with new revisions containing the following - changes: - - o The top level module description is updated to indicate that the - module conforms to the revised datastore architecture with a - combined config and state tree, and that the existing state tree - nodes are deprecated, to be obsoleted over time. - - o All status "current" data nodes under the existing "state" trees - are copied to the equivalent place under the "config" tree: - - * If a node with the same name and type already exists under the - equivalent path in the config tree then the nodes are merged - and the description updated. - - * If a node with the same name but different type exists under - the equivalent path in the config tree, then the module authors - must choose the appropriate mechanism to combine the config and - state nodes in a backwards compatible way based on the data - model design guidelines below. This may require the state node - to be added to the config tree with a modified name. This - scenario is expected to be relatively uncommon. - - * If no node with the same name and path already exists under the - config tree then the state node schema is copied verbatim into - the config tree. - - * As the state nodes are copied into the config trees, any - leafrefs that reference other nodes in the state tree are - adjusted to reference the equivalent path in the config tree. - - * All status "current" nodes under the existing "state" trees are - marked as "status" deprecated. - - o Augmentations are similarly handled to data nodes as described - above. - -C.2. Standardization of new YANG Data Models - - New standards based YANG modules, or those in active development, - should be designed to conform to the revised datastore architecture, - following the design guidelines described below, and only need to - provide combined config/state trees. - -Appendix D. Implications on other Documents - - The sections below describe the authors' thoughts on how various - other documents may be updated to support the datastore architecture - described in this document. They have been incorporated as an - appendix of this document to facilitate easier review, but the - expectation is that this work will be moved into another document as - soon as the appropriate working group decides to take on the work. - -D.1. Implications on YANG - - Note: This section describes the authors' thoughts on how YANG - [RFC7950] could be updated to support the datastore architecture - described in this document. It has been incorporated here as a - temporary measure to facilitate easier review, but the expectation is - that this work will be owned and standardized via the NETCONF working - group. - - o Some clarifications may be needed if this datastore model is - adopted. YANG currently describes validation in terms of the - configuration datastore while it really happens on the - configuration datastore. - -D.2. Implications on YANG Library - - Note: This section describes the authors' thoughts on how YANG - Library [RFC7895] could be updated to support the datastore - architecture described in this document. It has been incorporated - here as a temporary measure to facilitate easier review, but the - expectation is that this work will be owned and standardized via the - NETCONF working group. - - With the introduction of multiple datastores, it is important that a - server can advertise to clients which modules are supported in the - different datastores implemented by the server. In order to do this, - we propose that the "ietf-yang-module" ([RFC7895]) is revised, with - the following addition to the "module" list in the "module-list" - grouping: - - leaf-list datastore { - type identityref { - base ds:datastore; - } - description - "The datastores in which this module is supported."; - - } - -D.3. Implications to YANG Guidelines - - Note: This section describes the authors' thoughts on how Guidelines - for Authors and Reviewers of YANG Data Model Documents - [I-D.ietf-netmod-rfc6087bis] could be updated to support the - datastore architecture described in this document. It has been - incorporated here as a temporary measure to facilitate easier review, - but the expectation is that this work will be owned and standardized - via the NETCONF working group. - - It is important to design data models with clear semantics that work - equally well for instantiation in a configuration datastore and - instantiation in the datastore. - -D.3.1. Nodes with different config/state value sets - - There may be some differences in the value set of some nodes that are - used for both configuration and state. At this point of time, these - are considered to be rare cases that can be dealt with using - different nodes for the configured and state values. - -D.3.2. Auto-configured or Auto-negotiated Values - - Sometimes configuration leafs support special values that instruct - the system to automatically configure a value. An example is an MTU - that is configured to "auto" to let the system determine a suitable - MTU value. Another example is Ethernet auto-negotiation of link - speed. In such a situation, it is recommended to model this as two - separate leafs, one config true leaf for the input to the auto- - negotiation process, and one config false leaf for the output from - the process. - -D.4. Implications on NETCONF - - Note: This section describes the authors' thoughts on how NETCONF - [RFC6241] could be updated to support the datastore architecture - described in this document. It has been incorporated here as a - temporary measure to facilitate easier review, but the expectation is - that this work will be owned and standardized via the NETCONF working - group. - -D.4.1. Introduction - - The NETCONF protocol [RFC6241] defines a simple mechanism through - which a network device can be managed, configuration data information - can be retrieved, and new configuration data can be uploaded and - manipulated. - - NETCONF already has support for configuration datastores, but it does - not define an operational datastore. Instead, it provides the - operation that returns the contents of the datastore along - with all config false leaves. However, this operation is - incompatible with the new datastore architecture defined in this - document, and hence should be deprecated. - - There are two possible ways that NETCONF could be extended to support - the new architecture: Either as new optional capabilities extending - the current version of NETCONF (v1.1, [RFC6241]), or by defining a - new version of NETCONF. - - Many of the required additions are common to both approaches, and are - described below. A following section then describes the benefits of - defining a new NETCONF version, and the additional changes that would - entail. - -D.4.2. Overview of additions to NETCONF - - o A new "supported datastores" capability allows a device to list - all datastores it supports. Implementations can choose which - datastores they expose, but MUST at least expose both the - and datastores. They MAY expose - additional datastores, such as , , etc. - - o A new operation is introduced that allows the client to - return the contents of a datastore. For configuration datastores, - this operation returns the same data that would be returned by the - existing operation. - - o Some form of new filtering mechanism is required to allow the - device to filter the data based on the YANG metadata in addition - to other filters (such as the subtree filter). See also - Appendix E. - - o A new "with-metadata" capability allows a device to indicate that - it supports the capability of including YANG metadata annotations - in the responses to and requests. This is - achieved in a similar way to with-defaults [RFC6243], by - introducing a XML element to and - requests. - - * The capability would allow a device to indicate which types of - metadata are supported. - - * The XML element would specify which types of metadata are - included in the response. - - o The handling of defaults for the new configuration datastores is - as described in with-defaults [RFC6243], but that does not apply - for the operational state datastore that defines new semantics. - -D.4.2.1. Operational State Datastore Defaults Handling - - The normal semantics for the datastore are that all - values that match the default specified in the schema are included in - response to requests on the operational state datastore. This is - equivalent to the "report-all" mode of the with-defaults handling. - - The "metadata-filter" query parameter can be used to exclude nodes - with origin metadata matching "default", that would exclude nodes - that match the default value specified in the schema. - - If the server cannot return a value for any reason (e.g., the server - cannot determine the value, or the value that would be returned is - outside the allowed leaf value range) then the server can choose to - not return any value for a particular leaf, which MUST be interpreted - by the client as the value of that leaf not being known, rather than - implicitly having the default value. - -D.4.3. Overview of NETCONF version 2 - - This section describes NETCONF version 2, by explaining the - differences to NETCONF version 1.1. Where not explicitly specified, - the behavior of NETCONF version 2 is the same as for NETCONF version - 1.1 [RFC6241]. - -D.4.3.1. Benefits of defining a new NETCONF version - - Defining a new version of NETCONF (as opposed to extending NETCONF - version 1.1) has several benefits: - - o It allows for removal of the existing RPC operation, that - returns content from both the running configuration datastore - combined with all config false leaves. - - o It could allow the existing operation to also be - removed, replaced by the more generic that is named - appropriately to also apply to the operational datastore. - - o It makes it easier for clients and servers to know what reasonable - common baseline functionality to expect, rather than a collection - of capabilities that may not be implemented in a consistent - fashion. In particular, clients will able to assume support for - the datastore. - - o It can gracefully coexist with NETCONF v1.1. A server could - implement both versions. Existing YANG models exposing split - config/state trees could be exposed via NETCONF v1.1, whereas - combined config/state YANG models could be exposed via NETCONF v2, - providing a viable server upgrade path. - -D.4.3.2. Proposed changes for NETCONF v2 - - The differences between NETCONF v2 and NETCONF v1.1 can be summarized - as: - - o NETCONF v2 advertises a new base NETCONF capability - "urn:ietf:params:netconf:base:2.0". A server may advertise older - NETCONF versions as well, to allow a client to choose which - version to use. - - o NETCONF v2 removes support for the existing operation, that - is replaced by the on the operational datastore. - - o NETCONF v2 can publish a separate version of YANG library from a - NETCONF v1.1 implementation running on the same device, allowing - different versions of NETCONF to support a different set of YANG - modules. - -D.4.3.3. Possible Migration Paths - - A common approach in current data models is to have two separate - trees "/foo" and "/foo-state", where the former contains config true - nodes, and the latter config false nodes. A data model that is - designed for the revised architectural framework presented in this - document will have a single tree "/foo" with a combination of config - true and config false nodes. - - Two different migration strategies are considered: - -D.4.3.3.1. Migration Path using two instances of NETCONF - - If, for backwards compatability reasons, a server intends to support - both split config/state trees and the combined config/state trees - proposed in this architecture, then this can be achieved by having - the device support both NETCONF v1 and NETCONF v2 at the same time: - - o The NETCONF v1 implementation could support existing YANG module - revisions defined with split config/state trees. - - o The NETCONF v2 implementation could support different YANG - modules, or YANG module revisions, with combined config/state - trees. - - Clients can then decide on which type of models to use by expressing - the appropriate version of the base NETCONF capability during - capability exchange. - -D.4.3.3.2. Migration Path using a single instance of NETCONF - - The proposed strategy for updating existing published data models is - to publish new revisions with the state trees' nodes copied under the - config tree, and for the existing state trees to have all of their - nodes marked as deprecated. The expectation is that NETCONF servers - would use a combination of these updated models alongside new models - that only follow the new datastore architecture. - - o NETCONF servers can support clients that are not aware of the - revised datastore architecture, particularly if they continue to - support the deprecated operation: - - * For updated YANG modules they would see additional information - returned via the operation. - - * For new YANG modules, some of the state nodes may not be - available, i.e. for any state nodes that exist under a config - node that has not been configured (e.g., statistics under a - system created interface). - - o NETCONF servers can also support clients that are aware of the - revised datastores architecture: - - * For updated YANG modules they would see additional information - returned under the legacy state trees. This information can be - excluded using appropriate subtree filters. - - * New YANG modules, conforming to the datastores architecture, - would work exactly as expected. - -D.5. Implications on RESTCONF - - This section describes the authors' thoughts on how RESTCONF - [RFC8040] could be updated to support the datastore architecture - described in this document. It has been incorporated here as a - temporary measure to facilitate easier review, but the expectation is - that this work will be owned and standardized via the NETCONF working - group. - -D.5.1. Introduction - - RESTCONF [RFC8040] defines a protocol based on HTTP for configuring - data defined in YANG version 1 or 1.1, using a conceptual datastore - that is compatible with a server that implements NETCONF 1.1 - compliant datastores. - - The combined conceptual datastore defined in RESTCONF is incompatible - with the new datastore architecture defined in this document. There - are two possible ways that RESTCONF could be extended to support the - new architecture: Either as new optional capabilities extending the - existing RESTCONF RFC, or possibly as an new version of RESTCONF. - - Many of the required additions are common to both approaches, and are - described below. A following section then describes the potential - benefits of defining a new RESTCONF version, and the additional - changes that might entail. - -D.5.2. Overview of additions to RESTCONF - - o A new path {+restconf}/datastore//data/ to provide - a YANG data tree for each datastore that is exposed via RESTCONF. - - o Implementations can choose which datastores they expose, but MUST - at least expose both the and datastores. - They MAY expose the datastores as needed. - - o The same HTTP Methods supported on {+restconf}/data/ are also - supported on {+restconf}/datastore//data/ but - suitably constrained depending on whether the datastore can be - written to by the client, or is read-only. - - o The same query parameters supported on {+restconf}/data/ are also - support on {+restconf}/datastore//data/ except for - the following query parameters: - - o "metadata" - is a new optional query parameter that filters the - returned data based on the metadata annotation. - - o "with-metadata" - is a new optional query parameter that - indicating that the metadata annotations should be included in the - reply. - - o "with-defaults" is supported on all configuration datastores, but - is not supported on the operational state datastore path, because - it has different default handling semantics. - - o The handling of defaults (include the with-defaults query - parameter) for the new configuration datastores is the same as the - existing conceptual datastore, but does not apply for the - operational state datastore that defines new semantics. - -D.5.2.1. HTTP Methods - - All configuration datastores support all HTTP Methods. - - The datastore only supports the following HTTP methods: - OPTIONS, HEAD, GET, and POST to invoke an RFC operation. - -D.5.2.2. Query parameters - - [RFC7952] specifies how a YANG data tree can be annotated with - generic metadata information, that is used by this document to - annotate data nodes with origin information indicating the mechanism - by which the operational value came into effect. - - RESTCONF could be extended with an optional generic mechanism to - allow the filtering of nodes returned in a query based on metadata - annotations associated with the data node. - - RESTCONF could also be extended with an optional generic mechanism to - choose whether metadata annotations should be included in the - response, potentially filtering to a subset of annotations. E.g., - only include @origin metadata annotations, and not any others that - may be in use. - - Both of the generic mechanisms could be controlled by a new - capability. A new capability is defined to indicate whether a device - supports filtering on, or annotating responses with, the origin meta - data. - -D.5.2.3. Operational State Datastore Defaults Handling - - The normal semantics for the datastore are that all - values that match the default specified in the schema are included in - response to requests on the operational state datastore. This is - equivalent to the "report-all" mode of the with-defaults handling. - - The "metadata" query parameter can be used to exclude nodes with a - origin metadata matching "default", that would exclude (only config - true?) nodes that match the default value specified in the schema. - - If the server cannot return a value for any reason (e.g., the server - cannot determine the value, or the value that would be returned is - outside the allowed leaf value range) then the server can choose to - not return any value for a particular leaf, which MUST be interpreted - by the client as the value of that leaf not being known, rather than - implicitly having the default value. - -D.5.3. Overview of a possible new RESTCONF version - - This section describes a notional new RESTCONF version, by explaining - the differences to RESTCONF version 1. Where not explicitly - specified, the behavior of a new RESTCONF version is the same as for - RESTCONF version 1 [RFC8040]. - -D.5.3.1. Potential benefits of defining a new RESTCONF version - - Defining a new version of RESTCONF (as opposed to extending RESTCONF - version 1) has several potential benefits: - - o It could expose datastores, and models designed for the revised - datastore architecture, in a clean and consistent way. - - o It would allow the parts of RESTCONF that do not work well with - the revised datastore architecture to be omitted from the new - RESTCONF version. - - o It would make it easier for clients and servers to know what - reasonable common baseline functionality to expect, rather than a - collection of capabilities that may not be implemented in a - consistent fashion. - - o It could gracefully coexist with RESTCONF v1. A server could - implement both versions. Existing YANG models exposing split - config/state trees could be exposed via RESTCONF v1, whereas - combined config/state YANG models could be exposed via a new - RESTCONF version, providing a viable server upgrade path. - -D.5.3.2. Possible changes for a new RESTCONF version - - The differences between a notional new RESTCONF version and RESTCONF - version 1 (RESTCONF v1) [RFC8040] can be summarized as: - - o A new RESTCONF version would define a new root resource, and a - separate link relation in the /.well-known/host-meta resource. - - o A new RESTCONF version could remove support for the - {+restconf}/data path supported in RESTCONF v1. - - o A new RESTCONF version could publish a separate version of YANG - library from a RESTCONF v1 implementation running on the same - device, allowing different versions of RESTCONF to support a - different set of YANG modules. - -D.5.3.3. Possible Migration Path using a new RESTCONF version - - A common approach in current data models is to have two separate - trees "/foo" and "/foo-state", where the former contains config true - nodes, and the latter config false nodes. A data model that is - designed for the revised architectural framework presented in this - document will have a single tree "/foo" with a combination of config - true and config false nodes. - - If for backwards compatability reasons, a server intends to support - both split config/state trees, and the combined config/state trees - proposed in this architecture, then this could be achieved by having - the device support both RESTCONF v1 and the new RESTCONF version at - the same time: - - o The RESTCONF v1 implementation could support existing YANG module - revisions defined with split config/state trees. - - o The implementation of the new RESTCONF version could support - different YANG modules, or YANG module revisions, with combined - config/state trees. - - Clients can then decide on which type of models to use by choosing - whether to use the RESTCONF v1 root resource or the root resource - associated with the new RESTCONF version. - -Appendix E. Open Issues - - 1. NETCONF needs to be able to filter data based on the origin - metadata. Possibly this could be done as part of the - operation. - - 2. We need a means of inheriting @origin values, so whole - hierarchies can avoid the noise of repeating parent values. - Should "origin='system'" (or whatever we call it) be the default? - - 3. We need to discuss somewhere how remote procedure calls and - notifications/actions tie into datastores. RFC 7950 shows as an - example a ping action tied to an interface. Does this refer to - an interface defined in a configuration datastore? Or an - interface defined in the operational state datastore? Or the - applied configuration datastore? Similarly, RFC 7950 shows an - example of a link-failure notification; this likely applies - implicitly to the operational state datastore. The netconf- - config-change notification does explicitly identify a datastore. - I think we generally need to have remote procedure calls and - notifications be explicit about which datastores they apply to - and perhaps change the default xpath context from running plus - state to the operational state datastore. - Authors' Addresses Martin Bjorklund Tail-f Systems Email: mbj@tail-f.com Juergen Schoenwaelder Jacobs University