YANG Modules describing Capabilities for Systems and Datastore Update Notifications
Ericsson Magyar Tudosok korutja 111117 BudapestHungarybalazs.lengyel@ericsson.comFuturewei2330 Central ExpresswaySanta Clara, CA 95050USAludwig@clemm.orgHuaweibenoit.claise@huawei.com
OPS
NETCONF
This document defines two YANG modules,"ietf-system-capabilities” and
"ietf-notification-capabilities”.
The module "ietf-system-capabilities" provides a placeholder structure that
can be used to discover YANG related system capabilities for servers.
The module can be used to report capability information from the server
at run time or at implementation time, by making use of the YANG Instance Data
File Format.
The module "ietf-notification-capabilities" augments "ietf-system-capabilities"
to specify capabilities related to Subscription to YANG Notifications
for Datastore Updates.
Servers and/or publishers often have capabilities, values describing
operational behavior, that need to be conveyed to clients, which is
enabled by the YANG modules described in this document.
There is a need to publish this capability information as it is part of
the API contract between the server and client. Examples include maximum
size of data that can be stored or transferred, information about counters (whether a node supports "on-change" telemetry), etc. Such capabilities are often dependent on a vendor's implementation or the available resources at deployment.
Many such capabilities are specific to the complete system,
individual YANG datastores , specific parts of the YANG
schema, or even individual data nodes. It is a goal of this document to provide a
common way of representing such capabilities in a format that is:
vendor independentmachine-readableavailable in an identical format both at implementation time and at run time.
Implementation-time information is needed by Network Management System
(NMS) implementers.
An NMS implementation that supports notifications needs the information
about a system's capability so it can send "on-change" notifications.
If the information is not documented
in a way that is readily available to the NMS designer, but only as instance data from
the network node once it is deployed,
the NMS implementation will be delayed, because it has to wait for the
network node to be ready. In addition, the assumption that all
NMS implementers will
have a correctly configured network node available to retrieve data from
is an expensive proposition and may not always hold. (An NMS may need
to be able to handle many dozens of network node types.)
Often a fully functional NMS is a requirement for introducing a new
network node type
into a network, so delaying NMS readiness effectively also delays the
time at which a new network node type can be introduced into the network.
Implementation-time information is needed by system integrators.
When introducing a network node type into their network,
operators often need to integrate the node type into their own
management system. The NMS may have management functions that depend
on "on-change" notifications. The network operators need to plan their
management practices and NMS implementation before they decide to
buy the specific network node type. Moreover, the decision to buy the node
type sometimes depends on these management possibilities.
Run-time capability information is needed:
for any "purely model driven" application, e.g., a NETCONF-browser.
Such applications depend on reading models and capabilities at run time
to support all the publisher's available functionality.in case the capability might change during run time
e.g., due to licensing, HW constraints etc.to check that capability information provided earlier, at
implementation time, is what the publisher has implemented.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL
NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED",
"MAY", and "OPTIONAL" in this document are to be interpreted as
described in BCP 14
when, and only when, they appear in all capitals, as shown here.
The terms "YANG-Push", "on-change subscription" and "periodic
subscription" are used as defined in .
The terms "subscriber", "publisher" and "receiver" are used as
defined in .
The term "server" is used as defined in .
The terms "YANG instance data file format", "instance data", and
"instance data set" are used as defined in
.
In addition, this document defines the following terms:
"Implementation-time information": Information about the server's
behavior that is made available during the implementation of the server,
available from a source other than a running server.
"Run-time information": Information about the server's
behavior that is available from the running server via management protocols
such as NETCONF or RESTCONF .
Capability information is represented by instance-data based on one or
more "capability defining YANG modules". This allows a user to discover
capabilities both at implementation time and at run time.
For the implementation-time use case: Capabilities SHOULD be provided by the
implementer as YANG instance data files complying to
.
When provided, the file MUST be available already at implementation time,
retrievable in a way that does not depend
on a live network node. E.g., download from product website.
For the run-time use case: Capabilities SHOULD be available via
NETCONF or
RESTCONF from the live server (implementing
the publisher) during run time.
Implementations that support changing these capabilities at
run time SHOULD support "on-change" notifications about the
system-capabilities container.
The module "ietf-system-capabilities" provides a placeholder structure to be used to specify any YANG related system capability.
The module "ietf-notification-capabilities" is defined to allow a publisher to
specify capabilities related to "Subscription to YANG Notifications for Datastore Updates" , also known as YANG-Push, augmenting "ietf-system-capabilities".
The YANG data models in this document conform to the Network
Management Datastore Architecture (NMDA) defined in .
A specific case is the need to specify capabilities in the YANG-Push
functionality. As defined in a publisher may allow
subscribers to subscribe to updates from a datastore and will subsequently push
such update notifications to the receiver. Notifications may be sent
periodically or "on-change" (more or less immediately after each change).
A publisher supporting YANG-Push has a number of capabilities defined in
that are often determined during the implementation of the publisher. These include:
Supported (reporting) periods for "periodic" subscriptionsMaximum number of objects that can be sent in an updateThe set of datastores or data nodes for which "periodic" notification is supported.Additional capabilities if the optional "on-change" feature is supported include:
Supported dampening periods for "on-change" subscriptionsThe set of datastores or data nodes for which "on-change" notification is supported
Publishers might have some capabilities (or limitations) to document.
For example, how many update notifications and how many datastore node
updates they can send out in a certain time-period. Other publishers might
not support "periodic" subscriptions to all datastores. In some cases, a publisher supporting "on-change" notifications will not
be able to push updates for some object types "on-change". Reasons for
this might be that the value of the datastore node changes frequently
(e.g., in-octets counter), that small object changes are
frequent and irrelevant to the receiver (e.g., a temperature gauge changing 0.1
degrees within a predetermined and acceptable range), or
that the implementation is not capable of on-change notification for
a particular object. In all those cases, it will be important for
subscriber applications to have a way to identify which objects "on-change" notifications are supported and for which ones not.
Faced with the reality that support for "on-change" notification does not
mean that such notifications will be sent for any specific data node,
subscriber/management applications can not rely on the "on-change"
functionality unless the subscriber has some means to identify which objects
"on-change" notifications are supported for. YANG models are meant to be used as an
interface contract. Without identification of the data nodes actually supporting "on-change",
this contract would be incomplete.
Clients of a server (and subscribers to a publisher, as subscribers are also clients) need a method to gather capability information.
The module "ietf-system-capabilities" is defined to provide a
structure that can be used to discover (as read-only operational state) any YANG related system capability.This module itself does not contain any capabilities; it provides
augmentation points for capabilities to be defined in subsequent YANG
modules. The ietf-system-capabilies is used by other modules to augment in specific
capability information. Every set of such capabilities MUST be
wrapped in a container under the augment statement to cleanly
separate different groups of capabilities. These "wrapper containers" SHALL be
augmented in at /sysc:system-capabilities and /sysc:system-capabilities/sysc:datastore-capabilities/sysc:per-node-capabilities.
Capability values can be specified on system level,
datastore level (by selecting all nodes in the datastore)
or for specific data nodes of a specific datastore
(and their contained sub-trees).
Capability values specified for a specific datastore or
node-set override values specified on the system level.
Note: The solution is usable for both NMDA and non-NMDA systems.
For non-NMDA servers "config false" data is
considered as if it were part of the running datastore.
The following tree diagram
provides an overview of the data model.
This YANG module imports typedefs from and
a reference path from .
<CODE ENDS>
The YANG module "ietf-notification-capabilities" provides YANG-Push
related capability information.
The following tree diagram
provides an overview of the data model.
This YANG module imports a feature and typedefs from
and also imports the "ietf-system-capabilities" specified in this document.
<CODE ENDS>
The YANG modules specified in this document define a schema for data
that is designed to be accessed via network management protocols such
as NETCONF or RESTCONF
or as YANG instance data.
The lowest NETCONF layer is the secure transport layer, and the
mandatory-to-implement secure transport is Secure Shell (SSH)
. The lowest RESTCONF layer is HTTPS, and
the mandatory-to-implement secure transport is TLS .
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.
All protocol-accessible data nodes in augmented modules are read-only
and cannot be modified. The data in these modules is not security sensitive.
Access control may be configured, to avoid exposing the read-only data.
When that data is in file format, data should be protected against
modification or unauthorized access using normal file handling mechanisms.
The data in file format also inherits all the security considerations of
which has additional
considerations about read protections; and distinguishes between data at
rest and in motion.
This document registers two URIs in the IETF XML
registry . Following the format in
, the following registrations are
requested:This document registers two YANG modules in the
YANG Module Names registry .
Following the format in , the
following registrations are requested:For their valuable comments, discussions, and feedback, we wish to
acknowledge Andy Bierman, Juergen Schoenwaelder, Rob Wilton, Kent Watsen,
Eric Voit, Joe Clarke, Martin Bjorklund, Ladislav Lhotka, Qin Wu,
Mahesh Jethanandani, Ran Tao, Reshad Rahman and other members of the Netmod WG.The following examples use artwork folding
for better formatting.
The following instance data example describes the notification
capabilities of a hypothetical "acme-router". The router implements
the running and operational datastores.
Every change can be reported "on-change" from the running datastore,
but only "config false" nodes and some "config false" data from the
operational datastore.
Interface statistics are not reported "on-change", only two important counters.
Datastore subscription capabilities are not reported "on-change", as they
never change on the acme-router during run time.
The following examples use artwork folding
for better formatting.
The following instance data example describes the notification
capabilities of a hypothetical "acme-switch". The switch implements
the running, candidate and operational
datastores. Every change can be reported "on-change" from the
running datastore, nothing from the candidate datastore and all
"config false" data from the operational datastore.
"periodic" subscriptions are supported for running and
operational, but not for candidate datastores.
Note to RFC Editor (To be removed by RFC Editor)v18 - v19
IESG reviewv17 - v18
IESG reviewv16 - v17
AD review comments addressedv15 - v16
Two editorial comments from document shepherdv14 - v15
Address the last comments from document shepherdv13 - v14
Updated according to sheperds reviewAdded to import, which imported modules need to be implemented.Added notes to the RFC editorRe-arrange the sections, for a better reading flowMany editorial changesReplace YANG module prefixv12 - v13
Rearranged order of notification capability leafs into 3 groups:
generic, specific to periodic subscriptions, specific to on-change.Introduced artwork folding in the examplesUpdated to follow draft-ietf-netmod-yang-instance-file-format-10Various editing changesv11 - v12
Updated max-nodes-per-update descriptionReformatted YANG models with pyang -f yang --keep-comments --yang-line-length 69v10 - v11
Updated examplesUpdated typedef notification-supportv09 - v10
Removed description text from imports about the need for
implementing the imported module.Changed notification-support to bits with shorter namesAssigned enum values to supported-excluded-change-typeMade node-selector a choice to allow for future alternative
selection methods.Changed precedence of per-node-capabilities entries.
Precedence is now according to the position of entries in
the list.v08 - v09
Split the YANG module into two: ietf-system-capabilities
and ietf-notification-capabilities. Restructured/updated the draft
accordingly.v07 - v08
Prepared the YANG model to include other non-YANG-Push
related capabilities.Renamed the top level container to system-capabilitiesAdded a container subscription-capabilities to the grouping
subscription-capabilities to contain all subscription
related capabilitiesUpdated examples according to
draft-ietf-netmod-yang-instance-file-format-06.v06 - v07
Updated examples according to
draft-ietf-netmod-yang-instance-file-format-05.v05 - v06
Providing the capability data is only a "SHOULD" recommendation.
Some reviewers wanted MUST some wanted much less.The YANG module import statements now indicate the imported modules
that must be implemented not just available as import as requested by
the YangDoctors review.v04 - v05
Added new capabilities periodic-notifications-supported and
supported-excluded-change-type.Restructured YANG module to make the node-selector's usage similar
to how NACM uses it: "/" means the whole datastore.Small corrections, spelling, rewordingReplaced the term server with the term publisher except in cases
where we speak about datastores and functionality based on
get, getconfig operations. In this latter case it is really the
server functionality that is discussed.v03 - v04
Clarified recommended support for on-change notifications about
the datastore-subscription-capabilities.v02 - v03
Allow throughput related capabilities to be defined on top,
datastore or data node level. Described that specific capability
values always override generic ones.Indicate that non-NMDA servers can also use this model.Updated according to draft-ietf-netmod-yang-instance-file-format-04v01 - v02
Added instance data examplesOn-change capability can be defined per datastoreAdded "if-feature yp:on-change" where relevantUnified units usedv00 - v01
Add more capabilities: minimum period, supported period
max-number of objects, min dampening period,
dampening supported