| Internet-Draft | Network Models for IETF Network Slices | July 2023 |
| Barguil, et al. | Expires 11 January 2024 | [Page] |
This document exemplifies how the various data modules that are produced in the IETF can be combined in the context of IETF Network Slice Services delivery.¶
Specifically, this document describes the relationship between the IETF Network Slice Service models for requesting IETF Network Slice Services and both Service (e.g., the Layer-3 Service Model, the Layer-2 Service Model) and Network (e.g., the Layer-3 Network Model, the Layer-2 Network Model) models used during their realizations. In addition, this document describes the communication between an IETF Network Slice Controller (NSC) and the network controllers for the realization of IETF Network Slices.¶
The IETF Network Slice Service YANG model provides a customer-oriented view of the intended Network slice Service. Thus, once an NSC receives a request for a Slice Service request, the NSC has to map it to accomplish the specific objectives expected by the network controllers. Existing YANG network models are analyzed against the IETF Network Slice requirements, and the gaps in existing models are identified.¶
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 11 January 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.¶
The IETF has produced several YANG data models that are instrumental for automating the provisioning and delivery of connectivity services. An overview of these data models and a framework that describes how these various modules can be glued together are described in [RFC8969].¶
This document adopts the rationale of [RFC8969], but with a focus on the Network Slice Service [I-D.ietf-teas-ietf-network-slices].¶
For example, the IETF Network Slice Service YANG service model provides a customer-oriented view of the Network Slice Service. Once an IETF Network Slice controller (NSC) receives a Slice Service request, it needs to map it the underlying network capabilities to accomplish the intended service as expected by the network controller.¶
Several service models and network models, including the Layer-3 Service Model (L3SM) [RFC8049], the Layer-2 Service Model (L2SM) [RFC8466], and network models (e.g., the Layer-3 Network Model (L3NM) [RFC9182], the Layer-2 Network Model (L2NM) [RFC9291])) which may be utilized for the realizating of IETF Network Slice Services, are analyzed whether they can satisfy the IETF Network Slice requirements.¶
The document also identifies some gaps on existing models.¶
This document describes an architecture and a communication process between an NSC and other network controllers for IETF Network Slice Service management (creation, modification, etc.).¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
This document assumes that the reader is familiar with the contents of [RFC6241], [RFC7950], [RFC8309], and [I-D.ietf-teas-ietf-network-slices] as it uses terms from those RFCs.¶
This document uses the term "network model" as defined in Section 2.1 of [RFC8969].¶
As described in [I-D.ietf-teas-ietf-network-slices], the IETF Network Slice Controller (NSC) is a functional entity for the control and management of IETF Network Slices Services. As shown in {{fig1}}, an NSC exposes set of APIs for higher level systems to request an IETF Network Slice Service. These APIs can be used to manage other connecitivity services, such as managing the underlying delivery setup that that is required for the delivery of an IETF Network Slice Service. Such setup can be managed prior or during the process of an Network Service Slice. Concretely, the setup can be the management of bearers and attachement circuits that connect Service Demarcation Points (SDPs) to customer premises.¶
The NSC cutsomer-facing interface is invoked by a customer for managing an IETF Network Slice Service (i.e., creation, modification, or deletion). Upon receiving a request via a customer-facing interface, an NSC assesses whether it can satisify the request and then identifies the resources that are needed for realization of the IETF Network Slice Service. The network-facing interface is used to interact with one or more Network Controllers for the realization of the requested IETF Network Slice Service.¶
An NSC exposes a set of IETF service data models: mainly, the IETF Network Slice Service Interface [I-D.ietf-teas-ietf-network-slice-nbi-yang] or the Attachment Circuit-as-a-Service Interface [I-D.boro-opsawg-teas-attachment-circuit].¶
Network Controllers exposes to NSCs a set of network data models, such as the L3NM, the L2NM, or the Service Attachment Points (SAPs) [RFC9408]. Typically, by setting the service type to "network-slice", an NSC can retrieve via the the SAPs where Slice Services can be delivered to customers. Likewise, SAPs can also be used to retrieve where such services are delivered to customers through the network configuration described in the L3NM [RFC9182] or the L2NM [RFC9291]. These checks can be used as part of request feasibility checks.¶
This document focuses on how an NSC can be implemented in an operator's network.¶
+------------------------------------------+
| A higher level system |
| (e.g., E2E network slice orchestrator) |
+------------------------------------------+
A
| NSC Customer-facing APIs
V Customer Service Models
+------------------------------------------+
| IETF Network Slice Controller (NSC) |
+------------------------------------------+
A
| NSC Network-facing APIs
V Network Models
+------------------------------------------+
| Network Controller(s) |
+------------------------------------------+
¶
Figure 1: Network Slice Controllers as a Module of a Hierarchical SDN Controller.¶
Several architectural definitions have arisen on the IETF to support SDN and network slicing deployments. The architecture defined in [I-D.ietf-teas-ietf-network-slices] includes a three-level hierarchy.¶
depicts a possible architecture using similar concepts. It starts from a consumer or high-level operational systems. Then, the NSC function migth be part of a hierarchical network controller (e.g., as the MDSC in the ACTN context [RFC8453]) as a modular function. At the bottom, two network controllers, each one can handle multiple or single underlay technologies.¶
+------------------------------+
| High-level operation system. |
+--------------+---------------+
|IETF Network Slice Service Request
|
+-------------------v------------------+
| |
| Hierarchical Network |
| Controller/Orchestrator |
| |
| +-------------------------------+ |
| | IETF Network Slice Controller | |
| +-------------------------------+ |
| |
+-------------------+------------------+
|
|
+--------------+---------------+
| |
v v
+-------------+----------+ +-------------+----------+
| Network Controller | | Network Controller |
+-------------+----------+ +-------------+----------+
| |
| |
v v
Network Elements Network Elements
¶
Figure 2: IETF Network Slice Controller as a Module of a Hierarchical SDN Controller.¶
In other implementations, an NSC can be a standalone component that directly interact with a network controller, as depicted in {{fig3}}. In this scenario, a service request follows a "data-enrichment" path, where each entity adds more information to the service request.¶
This document describes how existing service models and network models interact to deliver a Network Slice Service in a service provider environment.¶
+-------------------------------+
| High-level operation system |
+-------------+-----------------+
|IETF Network Slice Request
|
+-------------v-----------------+
| IETF Network Slice Controller |
+-------------+-----------------+
|
|
+-------------v-----------------+
| Network Controller |
+-------------+-----------------+
|
|
v
Network Elements
¶
Figure 3: IETF Network Slice Controller as a Standalone Component¶
Alternatively, a NSC can be integrated with a network controller and directly realizes the Network Slice Services using device data models to configure the network devices. A sample architecture is depicted in {{fig4}}.¶
+-------------------------------+
| High-level operation system |
+-------------+-----------------+
|IETF Network Slice Request
|
+-------------v----------------+
| Network Controller |
| |
|+----------------------------+|
|| Network Slice Controller ||
|+----------------------------+|
| |
+-------------+----------------+
|
|
v
Network Elements
¶
Figure 4: IETF Network Slice Controller as a Module of a Network Controller.¶
The main requirements for an IETF Network Slice Service, based on the high-level slice requirements from multiple organizations and use cases are compiled in [I-D.ietf-teas-ietf-network-slice-use-cases]. To accomplish those requirements, a set of YANG data models have been proposed:¶
+---------------+
| Customer |
+-------+-------+
Customer Service Model |
e.g., slice-svc, ac-svc,| and bearer-svc
+-------+-------+
| Service |
| Orchestration |
+-------+-------+
Network Model |
e.g., l2vpn-ntw, l3vpn-ntw, sap, and | ac-ntw
+-------+-------+
| Network |
| Orchestration |
+-------+-------+
Network Configuration Model |
+-----------+-----------+
| |
+--------+------+ +--------+------+
| Domain | | Domain |
| Orchestration | | Orchestration |
+---+-----------+ +--------+------+
Device | | |
Configuration | | |
Model | | |
+----+----+ | |
| Config | | |
| Manager | | |
+----+----+ | |
| | |
| NETCONF/CLI..................
| | |
+--------------------------------+
+----+ Bearer | | Bearer +----+
|CE#1+--------+ Network +--------+CE#2|
+----+ AC | | AC +----+
+--------------------------------+
Site A Site B
¶
Figure 5: Overview of Data Models used for Network Slicing¶
This section outlines the compliance and operational aspects of Network Controller models with IETF Network slice requirements. Section presented the requirements of the IETF Network slice. In this subsection it is analyzed how available YANG models that can be used by a Network Controller can satisfy those requirements and identify gaps.¶
As per [I-D.ietf-teas-te-service-mapping-yang], Availability is a probabilistic measure of the length of time that a VPN/VN instance functions without a network failure. As per RFC 8330, The parameter "availability", as described in [G.827], [F.1703], and [P.530], is often used to describe the link capacity. The availability is a time scale, representing a proportion of the operating time that the requested bandwidth is ensured".¶
The calculation of the availability is not trivial and would need to be clearly scoped to avoid misunderstandings.¶
The set of Yang models proposed today allow to request tunnels/paths with different resiliency requirements in terms of protection and restoration. However, none of them include the possibility of requesting a specific availability (e.g. 99.9999%).¶
The LxVPN Models ([RFC9182] and [RFC9291]) allow to specify the bandwdidth at the interface level between the slice and the customer. In addition, the Service Mapping model [I-D.ietf-teas-te-service-mapping-yang] allows to bind a VPN to a given LSP, which have its bandwidth requirements. Additionally, TE models can force a give bandwidth in the connection between Provider Edges.¶
Previous comment applies to the incoming and outgoing bandwidth parameters required for the NFV-based services use case in [I-D.ietf-teas-ietf-network-slice-use-cases]. The Network sharing use case has Maximum and Guaranteed Bit Rate parameters. These parameters can be mapped to the TE tunnel models when setting up LSPs [I-D.ietf-teas-yang-te].¶
Protection schemes are mechanisms to define how to setup resources for a given connection. TE tunnel models [I-D.ietf-teas-yang-te] includes protection and restoration as two main attributes. The parameters included in the containers for protection and restoration cover the requirements of the IETF NS related with protection schemes. Similarly, TE models cover the parameter 'recovery time' for the network sharing use case.¶
Delay is a critical parameter for several IETF NS types. Every use-case defined in [I-D.ietf-teas-ietf-network-slice-use-cases] contains delay constraints. 5G use cases require 'delay tolerance', NFV-based services have the delay information within 'QoS metrics' and 'Bounded latency' in the network sharing use case.¶
During the realization of the IETF Network Slice, these parameters are part of the requirements of a TE tunnel configuration [I-D.ietf-teas-yang-te]. They can be included within the 'path-metric-bounds' parameter, so the created LSP fulfils the given metrics bounds like 'path-metric-delay-average' or 'path-metric-delay-minimum'.¶
The packet loss rate indicates the maximum rate for lost packets that the service tolerates in the link. During the realization of the IETF Network Slice, this attribute will influence the tunnel selection and the value is included in the [I-D.ietf-teas-yang-te] document as the 'path-metric-loss". The 'path-metric-loss' is a metric type, which measures the percentage of packet loss of all links traversed by a P2P path. This parameter is required for 5G services and network sharing use-case, while it is part of the 'QoS metrics' for the NFV-based services.¶
This section presents an initial analysis of the relationship between IETF NBI model parameters and L3SM and L2SM service model parameters.¶
The L3SM service parameters are defined in section 6.2 of [RFC8299].¶
The following parameters are considered, so far:¶
Bandwidth. This parameter indicates the bandwidth requirement between each CE and PE participating in the service, then referrign essentially to the required WAN link bandwidth. It is expressed in terms of bits per second and individually specified for both input and output. Despite it is not stated in RFC 8299, this parameter can be interpreted as the CIR/PIR expected for the CE - PE connection.¶
MTU. This parameter indicates the maximum PDU size expected for the layer-3 service. It is relevant since packets could be discarded in case the customer sends packets with longer MTU than the one expressed by this parameter.¶
QoS. Regarding QoS, two different kind of parameters are detailed.¶
QoS classification policy. This policy is used to classify the traffic received from the customer, and it is expressed as a set of ordered rules. It is used for marking the input traffic (from CE to PE) when the customer flows match any of the rules in the list, setting the appropriate target class of service (target-class-id).¶
QoS profile. This profile defines the traffic-scheduling to be applied to the flows for either Site-to-WAN, WAN-to-Site, or both directions. It contains the following information per class of service: rate-limit, latency, jitter and guaranteed bandwidth.¶
Multicast. This parameter identifies if the service is multicast, and if so, what is the role of the site in the customer multicast service topology (i.e., source, receiver, or both). It also defines the kind of multicast relationship with the customer (i.e., as a router requiring PIM, host requiring either IGMP or MLD, or both), as well as the support of IPv4, IPv6 or both.¶
Similarly L2SM model parameters are described in section 5.9 and 5.10 of [RFC8466].¶
Bandwidth. This parameter is related to the bandwidth between both CE and PE and can be expressed as CIR/EIR/PIR, in the ingress or egress direction, taking the CE as the point of reference.¶
MTU. This parameter refers to the maximum layer-2 PDU frame size.¶
QoS. The specification of the QoS follows a similar structure to the one described in the case of L3SM. Some differences apply, for instance, at the time of QoS classification, which is performed on top of layer-2 parameters (e.g., MAC addresses).¶
BUM traffic. This parameter allows to determine if a site acts as source, receiver, or both.¶
Availability. This parameter in the L2SM model relates to the capability of supporting multi-homing.¶
On the other hand, the IETF NS NBI YANG model supports a number of SLOs and SLEs in the form of network slice service policy attributes. Such policy can apply to per-network slice, per-connection group or per-connection indivudually (over-writting of attributes is allowed as more granular information is provided). The following SLO attributes are detailed:¶
One-way / Two-way bandwidth, indicating the guaranteed minimum bandwidth between any two NSEs (unidirectional / bidirectional).¶
One-way / Two-way latency, indicating the guaranteed minimum latency between any two NSEs (unidirectional / bidirectional).¶
One-way / Two-way delay variation, indicating the maximum permissible delay variation of the slice (unidirectional / bidirectional).¶
One-way / Two-way packet loss, indicating the maximum permissible packet loss rate between endpoints (unidirectional / bidirectional).¶
Additionally, the following SLEs are defined:¶
MTU, referring to the the maximum PDU size that the customer may use.¶
Security, indicating if encryption or other security measures are required between two endpoints.¶
Isolation, as a way of indicating the isolation level expected by the customer in the allocation of network resources.¶
Maximum occupancy level, to express the amount of flows to be admitted (and optionally a maximum number of countable resource units such as IP or MAC addresses).¶
Thus, an initial mapping between L3SM, L2SM and IETF NS NBI model can be performed as indicated in the follwoing table.¶
+-----------------------+-----------------------+--------------------------------+ | L3SM (RFC 8299) | L2SM (RFC 8466) | IETF NSC NBI YANG model | +-----------------------+-----------------------+--------------------------------+ | Bandwidth | Bandwidth (CIR, PIR) | Sum of bandwidth SLO per NSE | | | | counting all connections | +-----------------------+-----------------------+--------------------------------+ | MTU (layer 3 service) | MTU (layer 2 service) | MTU attribute in SLE | +-----------------------+-----------------------+--------------------------------+ | QoS | QoS | QoS | | ......................| ......................|................................| | - QoS classification | - QoS classification | Defined in the model as | | policy | policy | network-access-qos-policy-name | | | | to be applied per access-point | | ......................| ......................|................................| | - QoS profile | - QoS profile | | | - rate-limit | - rate-limit | Defined in the model as | | | | incoming/outgong rate-limits | | | | per end-point (or access-point)| | - latency | - latency | One-way / Two-way latency SLO | | - jitter | - jitter | One-way / Two-way delay | | | | variation SLO | | - bandwidth | - bandwidth | One-way / Two-way bandwidth SLO| +-----------------------+-----------------------+--------------------------------+ | Multicast | Broadcast, Unknown, | The need of replication can be | | | Unicast and Multicast | inferred from | | | (BUM) | ns-connectivity-type. Further | | | | details are not available (e.g.| | | | source or receiver role) | +-----------------------+-----------------------+--------------------------------+ | | Availability as dual | Availability as the ratio of | | | homing | up-time to | | | | total_time(up-time+down-time) | +-----------------------+-----------------------+--------------------------------+¶
{: #Table1 title='Mapping of IETF NS NBI and LxSM Service Attribute' artwork-align="center"}¶
The following consideration can be made:¶
While the QoS profile in the L3SM and the L2SM applies per service class, the parameters in the IETF Network Slcie Service Interface apply per connection. So if per- class granularity is required in an IETF network slice, then different connections have to be defined between the same end- points, one per service class.¶
A number of attributes are not defined in the L3SM nor the L2SM, such as packet loss, isolation, or security. Then, the L3SM and L2SM could not be sufficient to realize IETF Network Slice Services with such specific needs, unless those other objectives and expectations are provided by other means (e.g., realizing the L3SM thorugh technologies guaranteing dedicated resource allocation such as OTN).¶
This section presents an initial analysis of the relationship between IETF Network Slice Service model parameters and the L3NM and the L2NM parameters.¶
The L3NM service parameters are defined in Section 7.6.6 of [RFC9182].¶
As made in the previous section, some basic parameters are considered:¶
Bandwidth: The L3NM defines bandwidth in terms of the 'pe-to-ce-bandwidth' & 'ce-to-pe-bandwidth'. Both values are defined in absolute value in bps per interface. The model supports the usage of QoS policies to include inbound and outbound Rate limits.¶
MTU: The L3NM only supports the definition at the 'vpn-network-access' level.¶
QoS: The quality of service is differentiated in three-levels:¶
Multicast: mVPN is supported at vpn-node and vpn-network-access; Each level includes Rendezvous Point (RP), IGMP, PIM, and MLD definitions.¶
Similarly the L2NM parameters are described in Section 7.6.6 of [RFC9291]:¶
Bandwidth: The L2NM considers the same parameters 'pe-to-ce-bandwidth' & 'ce-to-pe-bandwidth'. However, per definition, the L2NM supports the differentiation of CIR, PIR values. It includes the same set of values described for the L2SM.¶
MTU: The L2NM differentiates among Service MTU and interface MTU. The MTU mismatch configuration is also supported as part of the 'vpn-service' configuration.¶
QoS: The quality of service is differentiated in two-levels:¶
Multicast: Discard options are available for unknown Broadcast, Unicast or Multicast (BUM).¶
Thus, an initial mapping between the L3NM, L2NM, and IETF Network Slice Service model can be performed as indicated in the follwoing table:¶
+-----------------------+-----------------------+--------------------------------+ | L3NM (RFC 9182) | L2NM (RFC 9291) | IETF NSC NBI YANG model | +-----------------------+-----------------------+--------------------------------+ | Bandwidth between CE | Bandwidth between CE | Sum of bandwidth SLO per NSE | | and PE. | and PE. Different | counting all connections | | | types: per CoS, per | | | | VPN network access, | | | | per site, etc. | | +-----------------------+-----------------------+--------------------------------+ | MTU (layer 3 service) | MTU (layer 2 service | MTU attribute in SLE | | | and link MTU) | | +-----------------------+-----------------------+--------------------------------+ | QoS | QoS | QoS | | ......................| ......................|................................| | - QoS classification | - QoS classification | Defined in the model as | | policy (based on | policy (based on | network-access-qos-policy-name | | layer 3 and 4 info)| layer 2 info) | to be applied per access-point | | ......................| ......................|................................| | - QoS profile (not | - QoS profile (not | Defined in the model as | | defined) | defined) | incoming/outgong rate-limits | | | | per end-point (or access-point)| | | | One-way / Two-way latency SLO | | | | One-way / Two-way delay | | | | variation SLO | | | | One-way / Two-way bandwidth SLO| +-----------------------+-----------------------+--------------------------------+ | Multicast | Broadcast, Unknown, | The need of replication can be | | | Unicast and Multicast | inferred from | | | (BUM) | ns-connectivity-type. Further | | | | details are not available (e.g.| | | | source or receiver role) | +-----------------------+-----------------------+--------------------------------+ | | | Availability as the ratio of | | N/A | N/A | up-time to | | | | total_time(up-time+down-time) | +-----------------------+-----------------------+--------------------------------+¶
{: #Table2 title='Mapping of IETF NS NBI and LxNM service attribute' artwork-align="center"}¶
An IETF Network Slice may use several underlying technologies. A new IETF Network Slice may be initiated following these steps:¶
Variations of this flow can be considered: * The customer requests bearers and attachment circuits, independent of any service that will be delivered over them. * The customer place a service-specific request with references to ACes. * The curstomer may update the bearers/AC/service delivery points during the lifetime of a service.¶
As a functional entity responsible for managing a network domain, a network controller can expose a set of YANG models to an NSC. An NSC can invokes these models during the realization of a IETF Network Slice Service. The following network models can be used for realization of IETF Network slices:¶
LxVPN network models:¶
Traffic Engineering models:¶
TE Service Mapping extensions:¶
ACLs and routing policies models:¶
The framework defined in [RFC8969] compiles a set of YANG data models for automating network services. The data models can be used during the service and network management life cycle (e.g., service instantiation, service provisioning, service optimization, service monitoring, service diagnosing, and service assurance). The so called Network models could be reused for the realization of Network slice requests.¶
The following models are examples of Network models that describe services.¶
TEAS has defined a collection of models to allow the management of Traffic Engineering tunnels.¶
The IETF has defined a YANG model to set up the procedure to map VPN service/network models to the TE models. This model, known as service mapping, allows the network controller to assign/retrieve transport resources allocated to specific services. At the moment there is just one service mapping model [I-D.ietf-teas-te-service-mapping-yang]. The "Traffic Engineering (TE) and Service Mapping Yang Model" augments the VPN service and network models.¶
This section does not intend to be prescriptive but descriptive about the potential usage of existing and proposed models for the provision of an IETF Network Slice service.¶
[I-D.contreras-teas-slice-controller-models] shows a potential internal structure of an IETF Network Slice Controller which can be divided into two components:¶
Note that this division in functional components of an IETF NSC is provided as an implementation option, not constraining any other implementation of functional structure.¶
Higher Level System
|
| NSC NBI
+-------------------------+
| NSC | |
| v |
| +-----------------+ |
| | | |
| | NS Mapper | |
| | | |
| +-----------------+ |
| | |
| v |
| +-----------------+ |
| | | |
| | NS Realizer | |
| | | |
| +-----------------+ |
| | |
+-------------------------+
| NSC SBI
v
Network Controllers
¶
Figure 8: IETF Network Slice Controller Structure¶
The details of IETF network slice mapper and realize are provided below for various implementation of NCS.¶
Referring to Figure 1 in an integrated architecture, an NCS is part of a Hierarchical SDN controller module, the NSC's and the Hierarchical Network Controller should share the same internal data and the same NBI. Thus, the H-SDN module must be able to:¶
Map: The NSC should process the customer request received through [I-D.ietf-teas-ietf-network-slice-nbi-yang]. The mapping process takes the network-slice SLOs selected by the customer selecting available Routing Policies and Forwarding policies for accomplishing those SLOs.¶
Realize: Create necessary network requests. The slice's realization can be translated into one or several LXNM Network requests, depending on the number of underlay controllers. Thus, the NSC must have a complete view of the network to map the orders and distribute them across domains. The realization should include the expansion/selection of Forwarding Policies, Routing Policies, VPN policies, and Underlay transport preference.¶
To maintain the data coherence between the control layers, the IETF Network Slice ID ns-id used of the [I-D.ietf-teas-ietf-network-slice-nbi-yang] must be directly mapped to the transport-instance-id at the VPN-Node level.¶
+
|
| IETF Network Slice Request:
draft-ietf-teas-ietf-network-slice-nbi-yang
| * network-slice-id
|
+-------------------v------------------+
| |
| Hierarchical Network |
| Controller/Orchestrator |
| |
| +-------------------------------+ |
| | IETF Network Slice Controller | |
| +-------------------------------+ |
| |
+-------------------+------------------+
IETF Network Slice Realizer: LXNM
VPN-id |
* transport-instance-id |
|
+--------------+---------------+
| |
v v
+-------------+----------+ +-------------+----------+
| Network Controller | | Network Controller |
+-------------+----------+ +-------------+----------+
| |
| |
v v
Network Elements Network Elements
¶
Figure 9 Workflow for the Slice Request in an Integrated Architecture.¶
Referring to Figure 2 when the Network Slice Controller is a stand-alone controller module, the NSC's should perform the same two tasks described in section 6.1:¶
Map: Process the customer request. The customer request can be sent using [I-D.ietf-teas-ietf-network-slice-nbi-yang]. The customer can also perform the network slice request using customized topologies.¶
Realize: Create necessary network requests. The slice's realization will be translated into one LXNM Network request. As the NCS has a topological view of the network, the realization can include the customer's traffic engineering transport preferences and policies.¶
+
|IETF Network Slice Request
draft-ietf-teas-ietf-network-slice-nbi-yang
* network-id
|
+-------------v-----------------+
| IETF Network Slice Controller |
+-------------+-----------------+
|
IETF Network Slice Realizer: LxNM
VPN-id |
* Underlay-transport
* transport-instance-id
|
+-------------v----------------+
| Network Controller |
+-------------+----------------+
|
|
v
Network Elements
¶
Figure 10 Workflow for the slice request in an stand-alone architecture.¶
The Network Slice Controller can be a module of the network controller. In that case, two options are available. One is to share the same device data model in the customer-facing and network-facing interfaces of the network controller. The direct translation would reduce the service logic implemented at the network controller level, grouping the mapping and translation into a single task:¶
+
| Slice Request based on
| Device Models
|
|
+------------------v------------------+
| |
| Network |
| Controller |
| |
| +------------------------------+ |
| | Network Slice Controller | |
| +------------------------------+ |
| |
+------------------+------------------+
| Device Models
|
v
Network Elements
¶
Figure 11 Workflow for the slice request in an stand-alone architecture.¶
A second option introduces a more complex logic in the network controller and creates an abstraction layer to process the transport slices. In that case, the controller should receive network slices creation requests and maintain the whole set of implemented slices:¶
+
|Slice Request
draft-ietf-teas-ietf-network-slice-nbi-yang
* network-id
|
|
+------------------v------------------+
| |
| Network |
| Controller |
| |
| +------------------------------+ |
| | Network Slice Controller | |
| +------------------------------+ |
| |
+------------------+------------------+
|
|
v
Network Elements
¶
Figure 12 Workflow for the slice request in an stand-alone architecture.¶
There are two main aspects to consider. On the one hand, the IETF Network Slice has a set of security related requirements, such as hard isolation of the slice, or encryption of the communications through the slice. All those requirements need to be analyzed in detailed and clearly mapped to the Network Controller and device interfaces.¶
On the other hand, the communication between the IETF network slicer and the network controller (or controllers or hierarchy of controllers) need to follow the same security considerations as with the network models.¶
The network YANG modules defines schemas for data that is designed to be accessed via network management protocols such as NETCONF [RFC6241] or RESTCONF [RFC8040].¶
The lowest NETCONF layer is the secure transport layer, and the mandatory-to-implement secure transport is Secure Shell (SSH) [RFC6242].¶
The lowest RESTCONF layer is HTTPS, and the mandatory-to-implement secure transport is TLS [RFC8466].¶
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.¶
The following summarizes the foreseen risks of using the Network Models to instantiate IETF network Slices:¶
This document is informational and does not require IANA allocations.¶
A wide variety of yang models are currently under definition in IETF that can be used by Network Controllers to instantiate IETF network slices. Some of the IETF slice requirements can be satisfied by multiple means, as there are multiple choices available. However, other requirements are still not covered by the existing models. A more detailed definition of those uncovered requirements would be needed. Finally a consensus on the set of models to be exposed by Network Controllers would facilitate the deployment of IETF network slices.¶