| Internet-Draft | Peer-Mount | October 2023 |
| Clemm, et al. | Expires 25 April 2024 | [Page] |
This document defines a mechanism, Peer-Mount, that allows YANG datastores to reference and incorporate information from remote datastores. This is accomplished by extending YANG with the ability to define mount points that reference data nodes in other YANG subtrees and subsequently allowing those data nodes to be accessed by client applications as if part of an alternative data hierarchy. In addition, means to manage and administer tho mount points are provided. This facilitates the development of applications that need to access network-wide data that treanscends individual network devices while ensuring network-wide data consistency. One example concerns example applications that require a network inventory and/or network topology with access to select management data within the nodes that comprise it.¶
The concept of Peer-Mount was first introduced in an earlier Internet Draft that was no longer pursued due to lack of interest at the time. It is being revived now in light of renewed IETF interest in network inventory, network topology, and related use cases, for which Peer-Mount is of specific interest. Other concepts defined in the earlier draft, notably Alias-Mount, are not considered here since they provide other capabilities that are less applicable to those topics.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 25 April 2024.¶
Copyright (c) 2023 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
This document introduces a new capability that allows YANG datastores [RFC7950] to incorporate and reference information from other YANG subtrees that reside on separate servers. The capability allows a client application to retrieve and have visibility of both local and remote YANG data as part of the same YANG tree accessed through a single server. This is provided by introducing a mountpoint concept. This concept allows to declare a YANG data node in a primary datastore to serve as a "mount point" under which a subtree with YANG data from another server can be mounted. To the client, this provides visibility to data from other subtrees, rendered in a way that makes it appear as if all of that data were an integral part of the same datastore. This enables users to retrieve local data as well as mounted data from remote in integrated fashion, using e.g. Netconf [RFC6241] or Restconf [RFC8040] [RFC8527] data retrieval primitives. Peer-Mount allows a server to effectively provide a federated datastore that includes YANG data from across the network.The concept is reminiscent of concepts in a Network File System that allows to mount remote folders and make them appear as if they were contained in the local file system of the user's machine.¶
Peer-Mount also takes inspiration from a new technique in data management known as data virtualization (https://en.wikipedia.org/wiki/Data_virtualization). Traditionally, data platforms like data lakes or data warehouses have relied on Extract-Transform-Load (ETL) pipelines in which data was ingested from sources and eventually, stored into the data platform for consumption. Data virtualization defines a new data access approach wherein data remains at its source and is collected and served on demand by the data platform only when a consumer requests such data. As a result, data is not duplicated in the data platform, but served directly to the consumer via the data platform. To this end, the data platform maintains virtual pointers to the source where the data con be retrieved.¶
This draft borrows heavily from an earlier draft [I-D.clemm-netmod-mount] in which the concept of Peer-Mount was first introduced. That draft had been accompanied by a second draft articulating the requirements to be addressed [I-D.voit-netmod-yang-mount-requirements]. Both drafts were eventually no longer pursued due to limited interest at the time.¶
Since then, things have changed in that use cases have emerged that would greatly benefit from Peer-Mount as a solution. This includes in particularly interest in developing models for network inventory as well as for network topologies. Both cases involve the need to provide a consolidated view of a network through a YANG data model that could be provided, for example, by a network controller. This interest has manifested itself in the creation of a new working group, Network Inventory YANG (ivy).¶
Peer-Mount can facilitate the development of network inventory as well as network topology models that allow to incorporate "live" management data from the network devices that make up the inventory. This can be achieved by mounting that data, for example aspects of their configuration or even current device state, below the network inventory and/or network topology entities. Benefits of this include the avoidance of data model redundancy (defining overlapping YANG data models both at the device and at the network inventory level), simplification of dealing with replicated data (single authoritative data ownership), and ultimately faster time to market and lower development cost. Other use cases to benefit from Peer-Mount include Digital Twin Networking and Digital Network Topology Maps, both of which require a holistic view of network inventory that includes live management data.¶
The earlier draft also included another variation of mount, Alias-Mount. Alias-Mount allowed for the definition of mountpoints that reference a local YANG subtree residing on the same server. That provided in effect an aliasing capability which provided for an alternative hierarchy and path to access the same YANG data. Alias-Mount could be thought of as a simpler version of Peer-Mount that does not specify a remote server. However, in the interest of simplicity, Alias-Mount is not included here as it does not contribute to the ability to provide a federated datastore providing a holistic network-wide view, which is the property that is of interest here.¶
Data that is mounted is authoritatively still owned by the server where the mounted data originates and resides on. That data is a part of that server's own datastore, regardless of whether or not it also happens to be mounted from a remote client somewhere else. This implies that from the view of the mounting system, there are a number of differences that apply to data that is mounted. Specifically, it means that the validation of integrity constraints is the responsibility of the authoritative owner, not of the server that is mounting that data as a client. Mounting does not impose additional constraints on the remote data; it merely provides a different view of the same data from remote.¶
The mountpoint concept applies in principle to operations beyond data retrieval, i.e. to configuration, RPCs, notification subscriptions [RFC8639], and YANG-Push subscriptions [RFC8641]. However, support for such operations involves additional considerations. Most significantly, in the case of configuration operations, additional considerations regarding transactions and locking would apply (which might now have to be supported across the network).¶
For this reason, in its initial version, only data retrieval operations (e.g. GET) will be supported for data that is mounted. Other operations that are directed at subtrees that include mounted information will simply be capped at the mountpoints, i.e. not be applied to mounted data.¶
It is conceivable that additional capabilities for operations on mounted information will be introduced at some point. However, to keep things simple, the specification of such capabilities is beyond the scope of this specification; they can be introduced incrementally over time and advertised by YANG servers through additional features at that time.¶
YANG does provide means by which modules that have been separately defined can reference and augment one another. YANG also does provide means to specify data nodes that reference other data nodes. However, all the data is assumed to be instantiated as part of the same datastore, for example a datastore provided through a NETCONF server. Existing YANG mechanisms do not account for the possibility that some information that needs to be referred not only resides in a different subtree of the same datastore, or was defined in a separate module that is also instantiated in the same datastore, but that is genuinely part of a different datastore that is provided by a different server.¶
The ability to mount information from local and remote datastores is new and not covered by existing YANG mechanisms. Until now, management information provided in a datastore has been intrinsically tied to the same server and to a single data hierarchy. In contrast, the capability introduced in this specification allows the server to present data that is instantiated on remote systems as if it were its own and contained in its own local data hierarchy.¶
The capability of allowing the mounting of information from other subtrees is accomplished by a set of YANG extensions that allow to define such mount points. For this purpose, a new YANG module is introduced. The module defines the YANG extensions, as well as a data model that can be used to manage the mountpoints and mounting process itself. Only the mounting module and its server (i.e. the "receivers" or "consumers" of the mounted information) need to be aware of the concepts introduced here. Mounting is transparent to the "providers" of the mounted information and models that are being mounted; any data nodes or subtrees within any YANG model can be mounted.¶
It should be mentioned that Peer-Mount is not to be confused with the ability to mount a schema, aka Schema Mount [RFC8528]. A Schema Mount allows to instantiate an existing model definition underneath a mount point which is then locally instantiated at that point. It does not allow to reference a set of YANG data that has already been instantiated somewhere else. In that sense, Schema-Mount resembles more a "grouping" concept that allows to reuse an existing definition in a new context, as opposed to referencing and incorporating existing instance information into a new context.¶
The ability to mount data from remote datastores is useful to address various problems that several categories of applications are faced with.¶
One category of applications that can leverage this capability are network controller applications that need to present a consolidated view of management information in datastores across a network. Generally speaking, applications may need to provide a network inventory [RFC8345] [I-D.wzwb-opsawg-network-inventory-management] which provides not only a list of inventory items in the network, but that also includes additional information about each of those items, such as their status or certain aspects of their configuration. Likewise, applications that provide a view of a network topology may want to include certain aspects about the status and other properties of nodes, termination points, and links that make up the topology. One example of such applications includes Network Digital Twins [I-D.irtf-nmrg-network-digital-twin-arch].¶
These applications are faced with the problem that in order to expose information, that information needs to be part of their own datastore. Today, this requires support of a corresponding YANG data module. In order to expose information that concerns other network elements, that information has to be replicated into the controller's own datastore in the form of data nodes that may mirror but are clearly distinct from corresponding data nodes in the network element's datastore. In addition, in many cases, a controller needs to impose its own hierarchy on the data that is different from the one that was defined as part of the original module. An example for this concerns interface data, both operational data (e.g. various types of interface statistics) and configuration data, such as defined in [RFC7223]. This data will be contained in a top-level container ("interfaces", in this particular case) in a network element datastore. The controller may need to provide its clients a view on interface data from multiple devices under its scope of control. One way of to do so would involve organizing the data in a list with separate list elements for each device. However, this in turn would require introduction of redundant YANG modules that effectively replicate the same interface data save for differences in hierarchy.¶
By directly mounting information from network element datastores, the controller does not need to replicate the same information from multiple datastores, nor does it need to re-define any network element and system-level abstractions to be able to put them in the context of network abstractions. Instead, the subtree of the remote system is attached to the local mount point. Operations that need to access data below the mount point are in effect transparently redirected to remote system, which is the authoritative owner of the data. The mounting system does not even necessarily need to be aware of the specific data in the remote subtree. Optionally, caching strategies can be employed in which the mounting system prefetches data.¶
A second category of applications concerns decentralized networking applications that require globally consistent configuration of parameters. When each network element maintains its own datastore with the same configurable settings, a single global change requires modifying the same information in many network elements across a network. In case of inconsistent configurations, network failures can result that are difficult to troubleshoot. In many cases, what is more desirable is the ability to configure such settings in a single place, then make them available to every network element. Today, this requires in general the introduction of specialized servers and configuration options outside the scope of NETCONF, such as RADIUS [RFC2866] or DHCP [RFC2131]. In order to address this within the scope of NETCONF and YANG, the same information would have to be redundantly modeled and maintained, representing operational data (mirroring some remote server) on some network elements and configuration data on a designated master. Either way, additional complexity ensues.¶
Instead of replicating the same global parameters across different datastores, the solution presented in this document allows a single copy to be maintained in a subtree of single datastore that is then mounted by every network element that requires awareness of these parameters. The global parameters can be hosted in a controller or a designated network element. This considerably simplifies the management of such parameters that need to be known across elements in a network and require global consistency.¶
It should be noted that for these and many other applications merely having a view of the remote information is sufficient. It allows to define consolidated views of information without the need for replicating data and models that have already been defined, to audit information, and to validate consistency of configurations across a network. Only retrieval operations are required; no operations that involve configuring remote data are involved.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, theyappear in all capitals, as shown here.¶
Data node: An instance of management information in a YANG datastore.¶
DHCP: Dynamic Host Configuration Protocol.¶
Datastore: A conceptual store of instantiated management information, with individual data items represented by data nodes which are arranged in hierarchical manner.¶
Data subtree: An instantiated data node and the data nodes that are hierarchically contained within it.¶
Mount client: The system at which the mount point resides, into which the remote subtree is mounted.¶
Mount point: A data node that receives the root node of the remote datastore being mounted.¶
Mount server: The server with which the mount client communicates and which provides the mount client with access to the mounted information. Can be used synonymously with mount target.¶
Mount target: A remote server whose datastore is being mounted.¶
NACM: NETCONF Access Control Model¶
NETCONF: Network Configuration Protocol¶
RADIUS: Remote Authentication Dial In User Service.¶
RPC: Remote Procedure Call¶
Remote datastore: A datastore residing at a remote node.¶
URI: Uniform Resource Identifier¶
YANG: A data definition language for NETCONF¶
YANG-Push: A mechanism that allows a client to subscribe to updates from a datastore, which are then automatically pushed by the server to the client.¶
The following example scenarios outline some of the ways in which the ability to mount YANG datastores can be applied. Other mount topologies can be conceived in addition to the ones presented here.¶
The need to maintain a network inventory is a requirement for many applications, for example applications that expect to operate on a network topology [RFC8345]. Network controllers can use the mounting capability as part of maintaining a network inventory and, more generally, presenting a consolidated view of management information across the network. This allows network controllers to expose network-wide abstractions, such as topologies or paths, multi-device abstractions, such as VRRP [RFC5798], and network-element specific abstractions, such as information about a network element's interfaces.¶
Without a mounting capability, a network controller would need to at least conceptually replicate data from network elements to provide such a view, incorporating network element information into its own controller model that is separate from the network element's, indicating that the information in the controller model is to be populated from network elements. This can introduce issues such as data inconsistency and staleness, in addition to operational overhead that is required to populate and sync that data. Equally important, it would lead to the need to define redundant data models: one model that is implemented by the network element itself, and another model to be implemented by the network controller. This leads to poor maintainability, as analogous information has to be redundantly defined and implemented across different data models. In general, controllers cannot simply support the same modules as their network elements for the same information because that information needs to be put into a different context. This leads to "node"-information that needs to be instantiated and indexed differently, because there are multiple instances across different data stores.¶
For example, a controller might want to maintain network inventory consisting of list of network elements. Underneath each network element, the network inventory should also contain respectrive system-level information. Without Peer-Mount, would require the definition of a YANG data model that defines the required system-level information as part of the network inventory, although the same information is also modeled as part of YANG data models that are instantiated at the respective network elements. The controller-level network inventory would require a separate data model (or set of data models) that repeats the same system-level information of the network element and which needs to be redundantly defined, implemented, and maintained. Any augmentations that add additional system-level information to the original module will likewise need to be redundantly defined, once for the YANG data model at the "system" level, a second time at the network inventory level.¶
By allowing a network controller (or other system maintaining a network inventory) to use Peer-Mount and directly mount information from network element datastores, the controller does not need to replicate the same information from multiple datastores. Perhaps even more importantly, the need to re-define any network element and system-level abstractions just to be able to put them in the context of network abstractions is avoided. In this solution, a network controller's datastore mounts information from many network element datastores. For example, the network controller datastore (the "primary" datastore) could implement a list in which each list element contains a mountpoint. Each mountpoint mounts a subtree from a different network element's datastore. The data from the mounted subtrees is then accessible to clients of the primary datastore using the usual data retrieval operations.¶
This scenario is depicted in Figure 1. In the figure, a Network Controller Datastore contains a network inventory rooted in Ninv. Ninv contains a list of nodes in the inventory, N11 and N12. M1 is the mountpoint for a subtree in the datastore in Network Element 1 and M2 is the mountpoint for a subtree in the datastore in Network Element 2. MDN1 is the mounted data node that is mounted from Network Element 1 below N11, and MDN2 is the data node mounted from Network Element 2 below N12.¶
+-------------+
| Network |
| Controller |
| Datastore |
| |
| +--Ninv |
| +--N11 |
| | +--M1*******************************
| +--N12 | *
| +--M2****** *
| | * *
+-------------+ * *
* +---------------+ * +---------------+
* | +--N1 | * | +--N5 |
* | +--N2 | * | +--N6 |
********> +--MDN2 | *********> +--MDN1 |
| +--N3 | | +--N7 |
| +--N4 | | +--N8 |
| | | |
| Network | | Network |
| Element | | Element |
| Datastore 1 | | Datastore 2 |
+---------------+ +---------------+
While network inventory serves as a primary motivator for the introduction of Peer-Mount, it can be used also for other applications. A second category of such applications concerns decentralized networking applications that require globally consistent configuration of parameters that need to be known across elements in a network. Today, the configuration of such parameters is generally performed on a per network element basis, which is not only redundant but, more importantly, error-prone. Inconsistent configurations lead to erroneous network behavior that can be challenging to troubleshoot.¶
Using the ability to mount information from remote datastores opens up a new possibility for managing such settings. Instead of replicating the same global parameters across different datastores, a single copy is maintained in a subtree of single datastore. This datastore can hosted in a controller or a designated network element. The subtree is subsequently mounted by every network element that requires access to these parameters.¶
In many ways, this category of applications is an inverse of the previous category: Whereas in the network controller case data from many different datastores would be mounted into the same datastore with multiple mountpoints, in this case many elements, each with their own datastore, mount the same remote datastore, which is then mounted by many different systems.¶
The scenario is depicted in Figure 2. In the figure, M1 is the mountpoint for the Network Controller datastore in Network Element 1 and M2 is the mountpoint for the Network Controller datastore in Network Element 2. MDN is the mounted data node (i.e. the root of the mounted subtree) in the Network Controller datastore that contains the data nodes that represent the shared configuration settings. (Note that there is no reason why the Network Controller Datastore in this figure could not simply reside on a network element itself; the division of responsibilities is a logical one.¶
+---------------+ +---------------+
| Network | | Network |
| Element | | Element |
| Datastore 1 | | Datastore 2 |
| | | |
| +--N1 | | +--N5 |
| | +--N2 | | | +--N6 |
| | +--N2 | | | +--N6 |
| | +--N3 | | | +--N7 |
| | +--N4 | | | +--N8 |
| | | | | |
| +--M1 | | +--M2 |
+-----*---------+ +-----*---------+
* * +---------------+
* * | |
* * | +--N10 |
* * | +--N11 |
*********************************************> +--MDN |
| +--N20 |
| +--N21 |
| ... |
| +--N22 |
| |
| Network |
| Controller |
| Datastore |
+---------------+
This section provides a rough illustration of the operations flow involving mounted datastores.¶
The first thing that should be noted about these operations flows concerns the fact that a mount client essentially constitutes a special management application that interacts with a subtree to render the data of that subtree as an alternative tree hierarchy. In Peer-Mount, the mount client constitutes in effect another application, with the remote system remaining the authoritative owner of the data. While it is conceivable that the remote system (or an application that proxies for the remote system) provides certain functionality to facilitate the specific needs of the mount client to make it more efficient, the fact that another system decides to expose a certain "view" of that data is fundamentally not the remote system's concern.¶
When a client application makes a request to a server that involves data that is mounted from a remote system, the server will effectively act as a proxy to the remote system on the client application's behalf. It will extract from the client application request the portion that involves the mounted subtree from the remote system. It will strip that portion of the local context, i.e. remove any local data paths and insert the data path of the mounted remote subtree, as appropriate. The server will then forward the transposed request to the remote system that is the authoritative owner of the mounted data, acting itself as a client to the remote server. Upon receiving the reply, the server will transpose the results into the local context as needed, for example map the data paths into the local data tree structure, and combine those results with the results of the remainder portion of the original request.¶
Data retrieval operations are the only category of operations that is supported for peer-mounted data. In that case, a Netconf "get" or "get-configuration" operation might be applied on a subtree whose scope includes a mount point. When resolving the mount point, the server issues its own "get" or "get-configuration" request against the remote system's subtree that is attached to the mount point. Any filters in the request are transposed to include only those portions that would be applicable to the remote subtree (in the process removing portions that would be applied locally above the mountpoint). The data that is returned is then inserted into the data structure that is in turn returned to the client that originally invoked the request.¶
The fact that data retrieval operations are the only category of operations that are supported for peer-mounted data does not preclude other operations to be applied to datastore subtrees that contain mountpoints and peer-mounted data. Peer-mounted data will simply be transparent to those operations. When an operation is applied to a subtree which includes mountpoints, mounted data is ignored for purposes of the operation. For example, for a Netconf "edit-config" operation that includes a subtree with a mountpoint, a server will ignore the data under the mountpoint and apply the operation only to the local configuration. Mounted data is treated as "read-only" data. The server does not even need to return an error message that the operation could not be applied to mounted data; the mountpoint is simply ignored.¶
In principle, it is conceivable that operations other than data-retrieval are applied to mounted data as well. For example, an operation to edit configuration information might expect edits to be applied to remote systems as part of the operation, where the edited subtree involves mounted information. However, editing of information and "writing through" to remote systems potentially involves significant complexity, particularly if transactions and locking across multiple configuration items across multiple remote systems are involved. Support for such operations will require additional capabilities, specification of which is beyond the scope of this specification.¶
Likewise, Peer-Mount does not extend towards RPCs that are defined as part of YANG modules whose contents is being mounted. Support for RPCs that involve mounted portions of the datastore, while conceivable, would require introduction of an additional capability, whose definition is outside the scope of this specification.¶
Finally, Peer-Mount does not extend towards notifications [RFC8639] nor YANG-Push [RFC8641]. However, it is conceivable and fairly straightforward to offer support for those operations in the future using a separate capability, definition of which is once again outside the scope of this specification.¶
Since mounting of data typically involves communication with a remote system, there is a possibility that the remote system will not respond within a certain amount of time, that connectivity is lost, or that other errors occur. Accordingly, the ability to mount datastores also involves mountpoint management, which includes the ability to configure timeouts, retries, and management of mountpoint state (including dynamic addition removal of mountpoints). Mountpoint management is discussed in section Section 6.3.¶
It is expected that some implementations will introduce caching schemes. Caching can increase performance and efficiency in certain scenarios (for example, in the case of data that is frequently read but that rarely changes), but increases implementation complexity. Caching is not required for Peer-Mount to work - in which case access to mounted data is "on-demand", in which the authoritative data node always gets accessed. Whether to perform caching is a local implementation decision.¶
When caching is supported by an implementation, it can benefit from the ability to subscribe to updates on remote data by remote servers. Some optimizations to facilitate caching support are discussed in section Section 8.5.¶
At the center of the module is a set of YANG extensions that allow to define a mountpoint in a YANG data model.¶
The first extension, "mountpoint", is used to declare a mountpoint. The extension takes the name of the mountpoint as an argument.¶
The second extension, "subtree", serves as substatement underneath a mountpoint statement. It takes an argument that defines the root node of the datastore subtree that is to be mounted, specified as string that contains a path expression. This extension is used to define mountpoints for Peer-Mount.¶
The third extension, "target", also serves as a substatement underneath a mountpoint statement. It takes an argument that identifies the target system from where a subtree is mounted. The argument is a reference to a data node that contains the information that is needed to identify and address a remote server, such as an IP address, a host name, or a URI [RFC3986]. It is conceivable that a mount point is contained in a container that is part of a list, with each list element containing a mount point instance that references a different target system, with target system information itself part of a separate list. The argument in this case will include the required index information to identify the list element which identifies the target system.¶
A mountpoint MUST be contained underneath a container, a list, or a case.¶
Only a single data node respectively subtree can be mounted at one time. While the mount target could refer to any data node, it is recommended that as a best practice, the mount target SHOULD refer to a container. It is possible to maintain e.g. a list of mount points, with each mount point each of which has a mount target an element of a remote list. However, to avoid unnecessary proliferation of the number of mount points and associated management overhead, when data from lists or leaf-lists is to be mounted, a container containing the list respectively leaf-list SHOULD be mounted instead of individual list elements.¶
It is possible for a mounted datastore to contain another mountpoint, thus leading to several levels of mount indirections. However, mountpoints MUST NOT introduce circular dependencies. In particular, a mounted datastore MUST NOT contain a mountpoint which specifies the mounting datastore as a target and a subtree which contains as root node a data node that in turn contains the original mountpoint. Whenever a mount operation is performed, this condition MUST be validated by the mount client.¶
YANG tree diagrams [RFC8340] have proven very useful to convey the "Big Picture". It would be useful to indicate in YANG tree diagrams where a given node serves as a mountpoint. We propose for this purpose also a corresponding extension to the structure representation convention. Specifically, we propose to prefix the name of the mounting data node with upper-case 'M'. The subtree being mounted is depicted with a "-->" and path to the subtree root. The identification of the target system is not depicted. The following diagram depicts a mountpoint "node-system-info" contained under data node "node", which contains also another data node "node-ID".¶
rw network
+-- rw nodes
+-- rw node [node-ID]
+-- rw node-ID
+-- M node-system-info --> /system
In addition to allowing to define mountpoints, the YANG module also contains facilities to manage the mountpoints themselves.¶
For this purpose, a list of the mountpoints is introduced. Each list element represents a single mountpoint. It includes an identification of the mount point, i.e. its location in the local datatree, and a mount target, i.e. the remote system hosting the remote datastore and a definition of the subtree of the remote data node being mounted. It also includes monitoring information about current status (indicating whether the mount has been successful and is operational, or whether an error condition applies such as the target being unreachable or referring to an invalid subtree).¶
In addition to the list of mountpoints, a set of global mount policy settings allows to set parameters such as mount retries and timeouts.¶
Each mountpoint list element also contains a set of the same configuration knobs, allowing administrators to override global mount policies and configure mount policies on a per-mountpoint basis if needed.¶
There are two ways how mounting occurs: automatic (dynamically performed as part of system operation) or manually (administered by a user or client application, mounted on request not in an arbitrary location but in a place that is permissible as per the model). A separate mountpoint-origin object is used to distinguish between manually configured and automatically populated mountpoints.¶
Whether mounting occurs automatically or is subject to management by a user or an application can depend on the mountpoint being defined, i.e. the semantics of the model.¶
When configured automatically, mountpoint information is automatically populated by the datastore that implements the mountpoint. The precise mechanisms for discovering mount targets and bootstrapping mount points are provided by the mount client infrastructure and outside the scope of this specification. Likewise, when a mountpoint should be deleted and when it should merely have its mount-status indicate that the target is unreachable is a system-specific implementation decision.¶
Manual mounting consists of two steps. In a first step, a mountpoint is manually configured by a user or client application through administrative action. Once a mountpoint has been configured, actual mounting occurs through an RPCs that is defined specifically for that purpose. To unmount, a separate RPC is invoked; mountpoint configuration information needs to be explicitly deleted. Manual mounting can also be used to override automatic mounting, for example to allow an administrator to set up or remove a mountpoint.¶
It should be noted that mountpoint management does not allow users to manually "extend" the model, i.e. simply add a subtree underneath some arbitrary data node into a datastore, without a supporting mountpoint defined in the model to support it. A mountpoint definition is a formal part of the model with well-defined semantics. Accordingly, mountpoint management does not allow users to dynamically "extend" the data model itself. It allows users to populate the datastore and mount structure within the confines of a model that has been defined prior.¶
The structure of the mountpoint management data model is depicted in the following figure as a YANG Tree Diagram [RFC8340].¶
module: ietf-peer-mount
+--rw mount-server-mgmt {mount-server-mgmt}?
+--rw mountpoints
| +--rw mountpoint* [mountpoint-id]
| +--rw mountpoint-id string
| +--ro mountpoint-origin? enumeration
| +--rw subtree-ref subtree-ref
| +--rw mount-target
| | +--rw (target-address-type)
| | +--:(IP)
| | | +--rw target-ip? inet:ip-address
| | +--:(URI)
| | | +--rw uri? inet:uri
| | +--:(host-name)
| | | +--rw hostname? inet:host
| | +--:(node-ID)
| | | +--rw node-info-ref? subtree-ref
| | +--:(other)
| | +--rw opaque-target-ID? string
| +--ro mount-status? mount-status
| +--rw manual-mount? empty
| +--rw retry-timer? uint16
| +--rw number-of-retries? uint8
+--rw global-mount-policies
+--rw manual-mount? empty
+--rw retry-timer? uint16
+--rw number-of-retries? uint8
<CODE BEGINS> file "ietf-peer-mount@20231023.yang"
module ietf-peer-mount {
namespace "urn:ietf:params:xml:ns:yang:ietf-peer-mount";
prefix pmt;
import ietf-inet-types {
prefix inet;
}
organization
"IETF NETMOD (NETCONF Data Modeling Language) Working Group";
contact
"WG Web: <http://tools.ietf.org/wg/netmod/>
WG List: <mailto:netmod@ietf.org>
WG Chair: Kent Watsen
<mailto:kwatsen@juniper.net>
WG Chair: Lou Berger
<mailto:lberger@labn.net>
Editor: Alexander Clemm
<mailto:ludwig@clemm.org>
Editor: Eric Voit
<mailto:evoit@cisco.com>
Editor: Aihua Guo
<mailto:aihuaguo.ietf@gmail.com>
Editor: Ignacio Dominguez Martinez-Casanueva
<mailto:ignacio.dominguezmartinez@telefonica.com>";
description
"This module provides a set of YANG extensions and definitions
that can be used to mount information from remote datastores.";
revision 2023-10-23 {
description
"Initial revision.";
reference
"draft-clemm-netmod-peermount-01.txt";
}
extension mountpoint {
argument name;
description
"This YANG extension is used to mount data from another
subtree in place of the node under which this YANG extension
statement is used.
This extension takes one argument which specifies the name
of the mountpoint.
This extension can occur as a substatement underneath a
container statement, a list statement, or a case statement.
As a best practice, it SHOULD occur as statement only
underneath a container statement, but it MAY also occur
underneath a list or a case statement.
The extension can take two parameters, target and subtree,
each defined as their own YANG extensions.
For Peer Mount, a mountpoint statement MUST contain both a
target and a subtree substatement for the mountpoint
definition to be valid.
The subtree SHOULD be specified in terms of a data node of
type 'pmt:subtree-ref'. The targeted data node MUST
represent a container.
The target system MAY be specified in terms of a data node
that uses the grouping 'pmt:mount-target'. However, it
can be specified also in terms of any other data node that
contains sufficient information to address the mount target,
such as an IP address, a host name, or a URI.
It is possible for the mounted subtree to in turn contain a
mountpoint. However, circular mount relationships MUST NOT
be introduced. For this reason, a mounted subtree MUST NOT
contain a mountpoint that refers back to the mounting system
with a mount target that directly or indirectly contains the
originating mountpoint.";
}
extension target {
argument target-name;
description
"This YANG extension is used to perform a Peer-Mount.
It is used to specify a remote target system from which to
mount a datastore subtree. This YANG
extension takes one argument which specifies the remote
system. In general, this argument will contain the name of
a data node that contains the remote system information. It
is recommended that the reference data node uses the
mount-target grouping that is defined further below in this
module.
This YANG extension can occur only as a substatement below
a mountpoint statement. It MUST NOT occur as a substatement
below any other YANG statement.";
}
extension subtree {
argument subtree-path;
description
"This YANG extension is used to specify a subtree in a
datastore that is to be mounted. This YANG extension takes
one argument which specifies the path to the root of the
subtree. The root of the subtree SHOULD represent an
instance of a YANG container. However, it MAY represent
also another data node.
This YANG extension can occur only as a substatement below
a mountpoint statement. It MUST NOT occur as a substatement
below any other YANG statement.";
}
feature mount-server-mgmt {
description
"Provide additional capabilities to manage remote mount
points";
}
typedef mount-status {
type enumeration {
enum "ok" {
description
"Mounted";
}
enum "no-target" {
description
"The argument of the mountpoint does not define a
target system";
}
enum "no-subtree" {
description
"The argument of the mountpoint does not define a
root of a subtree";
}
enum "target-unreachable" {
description
"The specified target system is currently
unreachable";
}
enum "mount-failure" {
description
"Any other mount failure";
}
enum "unmounted" {
description
"The specified mountpoint has been unmounted as the
result of a management operation";
}
}
description
"This type is used to represent the status of a
mountpoint.";
}
typedef subtree-ref {
type string;
description
"This string specifies a path to a datanode. It corresponds
to the path substatement of a leafref type statement. Its
syntax needs to conform to the corresponding subset of the
XPath abbreviated syntax. Contrary to a leafref type,
subtree-ref allows to refer to a node in a remote datastore.
Also, a subtree-ref refers only to a single node, not a list
of nodes.";
}
grouping mount-monitor {
description
"This grouping contains data nodes that indicate the
current status of a mount point.";
leaf mount-status {
type mount-status;
config false;
description
"Indicates whether a mountpoint has been successfully
mounted or whether some kind of fault condition is
present.";
}
}
grouping mount-target {
description
"This grouping contains data nodes that can be used to
identify a remote system from which to mount a datastore
subtree.";
container mount-target {
description
"A container is used to keep mount target information
together.";
choice target-address-type {
mandatory true;
description
"Allows to identify mount target in different ways,
i.e. using different types of addresses.";
case IP {
leaf target-ip {
type inet:ip-address;
description
"IP address identifying the mount target.";
}
}
case URI {
leaf uri {
type inet:uri;
description
"URI identifying the mount target";
}
}
case host-name {
leaf hostname {
type inet:host;
description
"Host name of mount target.";
}
}
case node-ID {
leaf node-info-ref {
type subtree-ref;
description
"Node identified by named subtree.";
}
}
case other {
leaf opaque-target-ID {
type string;
description
"Catch-all; could be used also for mounting
of data nodes that are local.";
}
}
}
}
}
grouping mount-policies {
description
"This grouping contains data nodes that allow to configure
policies associated with mountpoints.";
leaf manual-mount {
type empty;
description
"When present, a specified mountpoint is not
automatically mounted when the mount data node is
created, but needs to mounted via specific RPC
invocation.";
}
leaf retry-timer {
type uint16;
units "seconds";
description
"When specified, provides the period after which
mounting will be automatically reattempted in case of a
mount status of an unreachable target";
}
leaf number-of-retries {
type uint8;
description
"When specified, provides a limit for the number of
times for which retries will be automatically
attempted";
}
}
rpc mount {
description
"This RPC allows an application or administrative user to
perform a mount operation. If successful, it will result in
the creation of a new mountpoint.";
input {
leaf mountpoint-id {
type string {
length "1..32";
}
description
"Identifier for the mountpoint to be created.
The mountpoint-id needs to be unique;
if the mountpoint-id of an existing mountpoint is
chosen, an error is returned.";
}
uses mount-target;
leaf mountpoint {
type leafref;
description
"Identifies the data node to mount the target under."
}
output {
leaf mount-status {
type mount-status;
description
"Indicates if the mount operation was successful.";
}
}
}
rpc unmount {
description
"This RPC allows an application or administrative user to
unmount information from a remote datastore. If successful,
the corresponding mountpoint will be removed from the
datastore.";
input {
leaf mountpoint-id {
type string {
length "1..32";
}
description
"Identifies the mountpoint to be unmounted.";
}
}
output {
leaf mount-status {
type mount-status;
description
"Indicates if the unmount operation was successful.";
}
}
}
container mount-server-mgmt {
if-feature mount-server-mgmt;
description
"Contains information associated with managing the
mountpoints of a datastore.";
container mountpoints {
description
"Keep the mountpoint information consolidated
in one place.";
list mountpoint {
key "mountpoint-id";
description
"There can be multiple mountpoints.
Each mountpoint is represented by its own
list element.";
leaf mountpoint-id {
type string {
length "1..32";
}
description
"An identifier of the mountpoint.
RPC operations refer to the mountpoint
using this identifier.";
}
leaf mountpoint-origin {
type enumeration {
enum "client" {
description
"Mountpoint has been supplied and is
manually administered by a client";
}
enum "auto" {
description
"Mountpoint is automatically
administered by the server";
}
}
config false;
description
"This describes how the mountpoint came
into being.";
}
leaf subtree-ref {
type subtree-ref;
mandatory true;
description
"Identifies the root of the subtree in the
target system that is to be mounted.";
}
uses mount-target;
uses mount-monitor;
uses mount-policies;
}
}
container global-mount-policies {
description
"Provides mount policies applicable for all mountpoints,
unless overridden for a specific mountpoint.";
uses mount-policies;
}
}
}
<CODE ENDS>¶
Access to mounted information is subject to authorization rules. To the mounted system, a mounting client will in general appear like any other client. Authorization privileges for remote mounting clients need to be specified through NACM (NETCONF Access Control Model) [RFC8341].¶
It is conceivable to differentiate between different datastores on the remote server, that is, to designate the name of the actual datastore to mount, e.g. "running" or "startup". However, for the purposes of this spec, we assume that the datastore to be mounted is generally implied. Mounted information is treated as analogous to operational data; in general, this means the running or "effective" datastore is the target. That said, the information which targets to mount does constitute configuration and can hence be part of a startup or candidate datastore.¶
It is possible for the mounted subtree to in turn contain a mountpoint. However, circular mount relationships MUST NOT be introduced. For this reason, a mounted subtree MUST NOT contain a mountpoint that refers back to the mounting system with a mount target that directly or indirectly contains the originating mountpoint. As part of a mount operation, the mount points of the mounted system need to be checked accordingly.¶
It is possible that a mountpoint is broken. For example, a remote system could be unreachable due to many reasons, such as misconfiguration of the target system, communications failure, or administrative shutdown. When a mount client experiences such an issue, a retrieval operation will simply return the empty mountpoint (i.e., the data node representing the mountpoint without the mounted subtree underneath). The mount status can be retrieved separately if needed.¶
Under certain circumstances, it can be useful to maintain a cache of remote information. Instead of accessing the remote system, requests are served from a copy that is locally maintained. This is particularly advantageous in cases where data is slow changing, i.e. when there are many more "read" operations than changes to the underlying data node, and in cases when a significant delay were incurred when accessing the remote system, which might be prohibitive for certain applications. Examples of such applications are applications that involve real-time control loops requiring response times that are measured in milliseconds. However, as data nodes that are mounted from an authoritative datastore represent the "golden copy", it is important that any modifications are reflected as soon as they are made.¶
It is a local implementation decision of mount clients whether to cache information once it has been fetched. However, in order to support more powerful caching schemes, it becomes necessary for the mount server to "push" information proactively. For this purpose, it is useful for the mount client to subscribe for updates to the mounted information at the mount server. YANG-Push can be used for this purpose, creating for each mountpoint a subscription at the remote system for the mounted data and updating the local cache as updates are received.¶
It is conceivable to add a mechanism that allows to limit the data in a mounted subtree that would be returned as part of retrieval requests. This could be accomplished by specifying a filter expression as part of the mountpoint definition (for example via an additional substatement) or as part of the mountpoint instantiation (for example for manual mount operations via a separate RPC parameter). However, doing so would add significant complexity, requiring those filters to be specified as well as applied as part of proxy operations on top of any other filters. Users always have the option to specify their own subtree filter when requesting data retrieval, hence the only potential benefit of such a mechanism would lie in the simplification of caching implementations, limiting the amount of data to include in the cache. In the interest of keeping Peer-Mount simple, an additional filtering mechanism beyond that which is already supported by standard Netconf and RESTCONF operations is therefore not included.¶
Implementation specifics are outside the scope of this specification. That said, the following considerations apply:¶
Systems that wish to mount information from remote datastores need to implement a mount client. The mount client communicates with a remote system to access the remote datastore. To do so, there are several options:¶
The mount client acts as a NETCONF client to a remote system. To the remote system, the mount client constitutes essentially a client application like any other.¶
The mount client communicates with a remote mount server through a separate protocol or API.¶
It is the responsibility of the mount client to manage the association with the target system, e.g. to validate it is still reachable by maintaining a permanent association, perform reachability checks in case of a connectionless transport, etc.¶
It is the responsibility of the mount client to manage the mountpoints. This means that the mount client needs to populate the mountpoint monitoring information (e.g. keep mount-status up to data and determine in the case of automatic mounting when to add and remove mountpoint configuration). In the case of automatic mounting, the mount client also interacts with the mountpoint discovery and bootstrap process.¶
The mount client needs to also participate in servicing datastore operations involving mounted information. An operation requested involving a mountpoint is relayed by the mounting system's infrastructure to the mount client. For example, a request to retrieve information from a datastore leads to an invocation of an internal mount client API when a mount point is reached. The mount client then relays a corresponding operation to the remote datastore. It subsequently relays the result along with any responses back to the invoking infrastructure, which then merges the result (e.g. a retrieved subtree with the rest of the information that was retrieved) as needed. Relaying the result may involve the need to transpose error response codes in certain corner cases, e.g. when mounted information could not be reached due to loss of connectivity with the remote server, or when a configuration request failed due to validation error.¶
It is possible for a mount client to contain several mountpoints that each mount a different subtree from the same remote system. Implementations should consider maintaining a single management association (e.g., a single Netconf session) per target system, as opposed to maintaining a separate association for each mountpoint.¶
There is a certain amount of overhead associated with each mount point. The mount point needs to be managed and state maintained. Data subscriptions need to be maintained. Requests including mounted subtrees need to be decomposed and responses from multiple systems combined.¶
For those reasons, as a general best practice, models that make use of mount points SHOULD be defined in a way that minimizes the number of mountpoints required. Finely granular mounts, in which multiple mountpoints are maintained with the same remote system, each containing only very small data subtrees, SHOULD be avoided. For example, lists SHOULD only contain mountpoints when individual list elements are associated with different remote systems. To mount data from lists in remote datastores, a container node that contains all list elements SHOULD be mounted instead of mounting each list element individually. Likewise, instead of having mount points refer to nodes contained underneath choices, a mountpoint should refer to a container of the choice.¶
TBD¶
TBD¶
TBD¶
The following is a list of technical items for further discussion.¶
Get vs get-configuration. Should both get and get-configuration be supported as data retrieval operations, or get only? The reason for the distinction between get and get-configuration is generally ease of implementation and efficiency of the operation, simply returning contents from config file versus having to build that content from memory structures, hence having higher performance. Perhaps only "get" should be supported, with "get-config" ignoring any mounted information.¶
Target and Subtree YANG extension. Should target and/or subtree be mandatory statements? When RPCs to support manual mounting are supported, it is conceivable to allow for the manual mounting of any subtree from any remote system (as provided per a parameter of the RPC request), not just a specific subtree. In that case, any information mounted would be effectively treated analogous to anyxml - it will constitute just a blob. In that case, both "subtree" and "target" statements could be optional. On the other hand, including subtree as a statement will facilitate validation and letting client applications know what information to expect. Including target as a statement will facilitate system operation without needing to rely on manual mounting/unmounting. Removing target will faciliate implementation, as the system does not need to worry about automated mounting, with the system administrator in that case taking on the responsibility of applying the mount operations against the proper target.¶
RPC definitions and manual mount operations. Are manual mount operations really required or should they be removed? Also, check need for mountpoint ID and its use in mount/unmount operations.¶
Manual mount operations: If supported, need to decide whether to make manual mounting a one step or two step procedure. If done as two steps, a mountpoint would first be created (step 1), then the mount operation applied (step 2). An unpopulated mountpoint would in effect resemble a special container - it might be considered as a special type of container, where what is contained would be a remote subtree (populated by a second operation). This means an empty mountpoint would be considered as part of configuration. Alternatively, this could be done as one step - a mount operation instantiates the mountpoint and links it to the remote system in one step.¶
System-provided mountpoints. Clarify behavior and semantics of mountpoints that are automatically maintained by the system vs mountpoint managed via mount operations that are performed on request.¶
Mount point location. Would it simplify to only allow mountpoints below a container? This might facilitate the way in which remote systems are referenced, e.g. not needing to contain an index/key to a list. However, it would make paths longer than necessary and require introduction of more container objects (e.g. below list elements) than would otherwise be required. Containers can of course themselves always also be transitively contained underneath other data nodes such as lists.¶
Mount status. Define mount status and associated Finite State Machine.¶
Example. Provide examples for:¶
Mountpoint status. Decide whether to also define a piece of metadata to indicate mountpoint status that can be returned with the data node representing the mountpoint. It may make sense to do this to be able to distinguish the case when there is no data in the remote subtree from the case when there is an issue with the mountpoint status.¶
Filters on mounted data. Determine whether it would make sense to add a filter capability to reduce the amount of data retrieved as part of subtrees.¶
Local mountpoint. Peer-Mount mechanism could also become the means for combining data from multiple YANG modules implemented in the same datastore. This mechanism enables extending YANG modules with subtrees from other YANG modules without defining a new augmented YANG module and in cases where importing groupings or YANG schema mount cannot fit. For example, the Data Manifest [I-D.ietf-opsawg-collected-data-manifest] needs to reuse fragments of the YANG Library module, but since the available groupings cannot fit the needs of this use case, the Data Manifest copy-pasted fragments from YANG Library.¶