| Netmod Working Group | M. Betts, Ed. |
| Internet-Draft | ZTE |
| Intended status: Informational | N. Davis, Ed. |
| Expires: May 1, 2017 | Ciena |
| K. Lam, Ed. | |
| E. Varma, Ed. | |
| Nokia | |
| B. Zeuner, Ed. | |
| Deutsche Telekom | |
| S. Mansfield, Ed. | |
| Ericsson | |
| P. Doolan, Ed. | |
| Coriant | |
| October 28, 2016 |
Framework for Deriving Interface Data Schema from UML Information Models
draft-betts-netmod-framework-data-schema-uml-04
This draft describes a framework for how purpose and protocol specific interfaces can be systematically derived from an underlying common information model, focusing upon the networking and forwarding domain. The benefit of using such an approach in interface specification development is to promote convergence, interoperability, and efficiency.
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 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 May 1, 2017.
Copyright (c) 2016 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.
Interface specifications are often generated as point solutions where the designer codes a particular interface from domain (problem space) concepts that may not be explicitly captured, may be defined using localized terminology that is subject to ambiguity in interpretation, and is highly focused on a particular use-case/application. The designer typically provides a representation of the interface schema in the form of a data schema [RFC3444](i.e., data structures conveyed over the interface), which only exposes the view of the domain relevant at that specific interface. As this data schema is a simple statement of the particular interface, it solely describes relationships relevant to the specific realization, having no inherent relationship to other interfaces in the system.
Approaching the development of interface specifications on a per use-case/application basis tends to promote unnecessary variety through a proliferation of similar interfaces, resulting in unnecessary divergences that limit interoperability. It also risks confusion of representational artifacts with fundamental characteristics of the information to be conveyed across the interface. There is also a risk that conflicting representations of the same information may be generated. Finally, as each such interface appears to stand alone, it thereby fails to capture relationships with other aspects of the same (or different) domains that are not explicitly needed for the interface.
This draft describes a framework for how a protocol specific data schema and the encoding used for the interface can be systematically derived from an underlying common information model, focusing upon the networking and forwarding domain. The benefit of using such an approach in the development of interface specifications is to promote convergence, interoperability, and efficiency.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [RFC2119].
An information model condenses domain knowledge and insights to provide a representation of its essential concepts, structures, and inter-relationships. In capturing domain understanding, such a model offers a coherent and consistent terminology and structure, expresses the semantics of the domain, and interrelates all relevant aspects of the domain. It enables a consistent expression of information that improves interoperability between software components at interfaces derived from it. A "good" information model should capture domain best practices, and be designed to support domain variety as well as extensibility and evolution. Examples of domains include networking and forwarding, storage, etc. A common industry information model is the assembly of all domain information models, which inter-relate at "touch points". Note that a common industry information model should not be interpreted as being a monolithic entity; in particular, a modular structure is essential to allow for extensibility.
There may be several relevant views of any particular domain, depending upon the perspective of the viewer, all of which are interrelated and involve subsets of the information model, and none of which contradict each other. (It should be noted that one view provides the information model representation of the overall domain.) To form a particular (purpose-specific) view, some elements of the model may be pruned. Additionally, for efficiency, some systematic refactoring of the information model may also occur.
In this draft, the term data schema is used in the context of either: (i) a specific protocol that is used to implement a purpose specific interface, or (ii) a programming language that is used to invoke a purpose specific API. Note that it is possible to map directly from the purpose specific information model to interface encoding.
While a purpose specific interface/API is not a simple direct encoding of the information model of the overall domain, it is by its nature based on a relevant view of the information model of the domain (i.e., a purpose specific information model view). It must be completely and consistently traceable to this view and should use the associated domain terminology. Depending on its application, a particular view may lead to a number of encoded forms at various types of interfaces/APIs. The information model does not dictate the encoded form, which will depend upon such factors as necessary capability, interaction style, and programming language.
This section introduces the Unified Modeling Language (UML), which has been used to model application structure, behavior, and architecture (in addition to business process and data structure). It also provides references to existing and ongoing work on standard information models based on UML.
The information model is expressed in terms of the Unified Modeling Language (UML) [OMG_UML], which was developed by the Object Management Group. It is a general-purpose modeling language in the field of software engineering. In 2000 the Unified Modeling Language was also accepted by the International Organization for Standardization (ISO) as an approved ISO standard [ISO_IEC_UML]. UML may be used in four ways:
UML defines a number of basic model elements (UML artifacts), such as object classes, attributes, associations, interfaces, operations, operation parameters, data types, etc. In order to assure a consistent and harmonized modelling approach, and to ensure uniformity in the application of UML to a problem domain, a subset of the basic model artifacts should be selected according to guidelines for creating an information model expressed in UML [ONF_TR-514]. The guidelines are generic; i.e., they are not specific to any particular domain that the information model is addressing, nor are they restricted to any particular protocol interface data schema. A UML information model may be created using Open Source UML tools; guidelines to be taken into account during the creation of a UML information model for the Open Source tool Papyrus have been developed in [ONF_TR-515].
Information models expressed in UML, primarily focused upon the networking and forwarding domain, have been, and are in the process of being, developed in ITU-T, TM Forum, NGMN, 3GPP, MEF, ONF, and others.
ONF has defined the Core Model of the ONF Common Information Model (ONF-CIM). The ONF Core Model [ONF_TR-512] provides a representation of network resources for the purpose of management-control and is independent of specific forwarding technology. The Core Model can be augmented to provide forwarding technology specific representation.
ITU-T Recommendations are focused on understanding the telecommunications problem space and developing information models addressing network and network element considerations. Some examples of available standard ITU-T information models relevant to the networking and forwarding domain include:
The above information models are developed from ITU-T Recommendations that define the respective transport technology functional models and management requirements.
The TM Forum community has likewise developed extensive models of the same space from the network level management perspective [TMF_MTNM] [TMF_MTOSI] [TMF_TR225]. The basis for all functions made available to the network level management is defined in the protocol-neutral network element level management work done in ITU-T. Its models thus complement the ITU-T information models. In further collaboration with 3GPP, considerable joint effort has been devoted to develop a consistent and coherent approach to that space. Most recently (September 2016), a Collaboration Agreement was signed between the MEF Forum, ONF, and TM Forum to enable common model collaboration on Information Model constructs and network resource Information Model.
The NGMN has published a document called Next Generation Converged Operations Requirements (NGCOR) [NGMN_NGCOR], with the expressed purpose of taking these requirements into account when converged management interfaces for mobile and fixed networks are being standardized in the SDOs. An ongoing collaboration called the Multi-SDO Project on Converged Management is taking care that the requirements are considered during the specification of new interfaces. It includes participants from ETSI, NGMN, TMF, 3GPP, and other SDOs, equipment vendors, OS vendors and service providers.
This section outlines the overall structure of a modular and evolvable common information model and how purpose specific IM views and data schema may be derived from it [ONF_TR-513].
As illustrated in Figure 1 below, the common information model is comprised of a library of model artifacts (objects, attributes, and associations) organized into a number of sub-models, to facilitate the independent development of technology and application specific extensions. The Core Model refers to information model artifacts that are intended for use by multiple applications and/or forwarding technologies. For purposes of navigability, the Core Model is further sub-structured into Core Network Model (CNM), Core Foundation Model, Core Physical Model, and the Core Specification Model (these are further discussed in Section 4.2.1). The forwarding technology specific model refers to technology specific extensions; e.g., for OTN, Ethernet, MPLS-TP, SDH, etc. The application specific model refers to extensions for supporting particular applications.
+-------------+
| Common |
| Information |
| Model |
| (CIM) |
|+-----------+|
||Core Model ||
||(TR-512) ||
||* CNM ||
||* Foundat. ||
||* Physical ||
||* ... ||
||* ... || +----------+ +---------+ +---------+
|+-----------+| | | Map |Interface| Map |Interface|
|+-----------+| | View |---+\| 1 data |---+\| 1 |
||Technology ||-------\ | of |---+/| schema |---+/|encoding |
||specific ||Prune/ \| CIM | +---------+ +---------+
||models ||refactor/| for a |
|+-----------+|-------/ |particular| Map +---------+
|+-----------+| . | purpose |-------------------+\|Interface|
||Compute || . | |-------------------+/| 2 |
||specific ||-------\ +----------+| . . |encoding |
||models ||Prune/ \ +--.--.----+| . . +---------+
|+-----------+|refactor/ +-.--.-----+|. .
| . |-------/ . . . .
| . | . . . . .
| . | . . . . .
|+-----------+| . . . . .
||Storage ||. . . . . .
||specific || . . . . . .
||models || . .. . . .
|+-----------+| . . . . .
| | . .. . . .
+-------------+ . . .. . .
. . . . . . .
. . . . .. . .
. . . . . . . .
+-----------.-----.-----.------.--.---------------.---.------------
|Guidelines . . . . . . . . \
|+----------\ +------\ +------\ +----------\ +------------ |
|| TR-513 \ |TR-514 \ |TR-515 \ | TR-513 \ +------------\\ |
|| Model | |Use of | |Papyrus| | Common | | Interfac |||
|| structure | |UML | |GitHub | | process | | specific |+|
|+-----------+ +-------+ +-------+ +-----------+ +-------------+ |
+------------------------------------------------------------------+
High-level common information model structure and methodology for deriving interface protocol specific data schema/interface encodings
Figure 1
The following subsections provide further elaboration of the high-level methodology introduced above.
As introduced earlier, a common information model includes the objects/packages, their properties (represented as attributes), and their relationships, etc. that are necessary to describe the domain for the applications being developed. It will be necessary to continually expand and refine the common model over time as new forwarding technologies, capabilities and applications are encompassed and new insights are gained. To allow these extensions to be made in a seamless manner, the common information model is structured into a number of sub-models. This modelling approach enables application specific and forwarding technology specific extensions to be developed by domain experts with appropriate independence.
Over time, some part(s) of the common information model may need to be augmented or changed. Any such areas are clearly identified using model lifecycle stereotypes (controlled annotations; e.g., experimental, preliminary, obsolete) to ensure ongoing compatibility and to ease migration. The use of the lifecycle stereotypes is described in the UML modeling guidelines [ONF_TR-514].
The core model is organized into a number of sub-models, each addressing a specific topic to allow for easier navigation. Currently, these consist of the Core Network Model (CNM), Core Foundation Model, Core Physical Model, and the Core Specification Model [I-D.lam-topology].
These sub-models contain the artifacts (objects, attributes and associations) that relate solely the specific technology or application. In some cases, the addition of an application or technology sub-model will also require, and result in, enhancement of the core model.
The next step is the development of a purpose specific information model, which is a true subset of the common information model. A purpose specific information model will typically be much smaller than the entire common information model. To provide maximal reuse, the purpose specific view is developed in two steps: (1) prune and refactor a copy of the artifacts of the common information model to provide a model of the network to provide a purpose specific information model of the network to be managed, where only those artefacts that represent the capabilities that are both in scope and supported are included, and (2) define the access rights for the various groups of users that will manage that network. Pruning and refactoring provides a purpose specific information model that represents the capabilities of the network of interest. The definition of access rights provides the ability to limit the actions that can be taken by the various user groups that will use that information model.
A data schema (DS) is developed in the context of either a specific protocol that is used to implement a purpose specific interface or a programming language that is used to invoke a purpose specific API. The DS is constructed by mapping of the purpose specific information model together with the operations patterns from the common information model to provide the interface protocol specific DS that includes operations and notifications. The operations should include data structures taken directly from the purpose specific information model view with no further adjustment.
The development of the data schema should consider the following:
This step encodes either a purpose specific data schema or a purpose specific information model into either: (i) a specific protocol that is used to implement a purpose specific interface, or (ii) a programming language that is used to invoke a purpose specific API. If the interface is encoded directly from the purpose specific information model then the interface operations must be added as described above.
Applying the methodology outlined in Section 4, protocol-specific interface data schema/encodings may be derived from existing, and emerging, standard UML information models addressing the forwarding and networking domains (e.g., [ITU-T_G.7711], G.874.1 [ITU-T_G.874.1]).
In order to assure a consistent and valid data modelling language representation that enables maximum interoperability, translation guidelines from UML information models to data schema/interface encodings are required. A set of translation rules also assists in development of automated tooling.
Guidelines have been developed for translation of data modeled with UML to YANG including mapping of object classes, attributes, data types, associations, interfaces, operations and operation parameters, notifications, and lifecycle [ONF_TR-531], [I-D.mansfield-uml].
It should be noted that the concept of deriving protocol-specific modules from UML information models is not new (e.g., MEF 38 [MEF_38] and MEF 39 [MEF_39] provide YANG modules derived from UML information models G.8052 [ITU-T_G.8052] and MEF 7.1 [MEF_7.1] for Service OAM Fault and Performance Monitoring, respectively.). What is new is the concept of an open, modular, evolvable common information model, coupled with an associated suite of essential guidelines and tooling (e.g., UML, Open Source tooling, translation, etc.), for realizing a coherent set of solution modules.
This draft describes a modular and scalable approach for systematically deriving purpose and protocol specific interfaces from an underlying common information model, focusing upon the networking and forwarding domain. Building upon an underlying common information modeling description of network resources (functionality, capabilities, flexibility) is a key enabler to convergence and interoperability. It is also future proof in the sense that the emergence of new protocols becomes solely a non-disruptive mapping issue. It should be noted that not all domains require development of information model prior to solutions development; the domains where this is of greatest benefit involve networking domains requiring support for an enhanced level of control and network programmability.
Dave Hood
Ericsson
USA
email dave.hood@ericsson.com
This memo includes no request to IANA.
This informational document only describes a framework for deriving interface data schema from UML Information Models. As such, security concerns are out of the scope of this document.
| [RFC2119] | Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, March 1997. |
| [RFC3444] | Pras, A. and J. Schoenwaelder, "On the Difference between Information Models and Data Models", RFC 3444, DOI 10.17487/RFC3444, January 2003. |
TBD