A YANG Data Model for Network Resource Partitions
(NRPs)Huawei Technologies101 Software Avenue, Yuhua DistrictNanjingJiangsu210012Chinalana.wubo@huawei.comHuawei TechnologiesDivyashree Techno ParkBangaloreKarnataka560066Indiadhruv.ietf@gmail.comOrangeRennes 35000Francemohamed.boucadair@orange.comChina UnicomBeijingChinachengying10@chinaunicom.cnChina MobileBeijingChinagongliyan@chinamobile.com
Routing Area
This document defines a YANG data model of Network Resource Partition
(NRP) for the NRP management operation. The model can be used for the
realization of IETF Network Slice Services. defines IETF
Network Slice services that provide connectivity coupled with network
resources commitment between a number of Service Demarcation Points
(SDPs) over a shared network infrastructure and, for scalability and
agility concerns, defines Network Resource Partition (NRP) to host one
or a group of network slice services according to characteristics
including Service Level Objectives (SLOs) and Service Level Expectations
(SLEs). analyzes the
scalability issues of network slice services in detail and suggests
candidate technologies of control and forwarding planes of the NRP.This document defines a YANG module of NRP that the IETF NSC (Network
Slice controller) can use to manage NRP instances to realize the network
slicing services. According to the YANG model classification of , the NRP model is a network configuration model.The following terms are defined in and are
used in this specification: configuration datastate dataThe following terms are defined in and are
used in this specification: augmentdata modeldata nodeThe terminology for describing YANG data models is found in
.The tree diagram used in this document follows the notation defined
in . section 6.1
introduces the concept of NRP, which is a collection of resources
(bufferage, queuing, scheduling, etc.) in the underlay network to
provide specific SLOs and SLEs for connectivity constructs of IETF
Network Slice services.
provides some solutions to realize network slicing in IP/MPLS networks.
Additionally
provides analysis and possible optimizations of the control plane and
data plane of NRP in IP/MPLS networks for better scalability. The
following are some common NRP attributes for NRP management identified
based on the analysis:NRP instantiationNRP partition type: Refers to various NRP resource partition
methods, such as control plane partition, data plane partition,
or hybrid partition, etc.NRP topology generation method: Topologies can be created
using multiple methods. For example, NRP links can be all links
in the underlay topology, or explicitly selected links from the
topology or implicitly selected from the various existing
topologies.NRP resource reservation: Reserves link resources for the
NRP, including bandwidth, queuing, and other resource
partitioning.NRP control plane: Mechanisms that provide routing and
forwarding to one or a group of network slice traffic to ensure
the corresponding SLO and SLE through NRP link resources, e.g.
distributed control plane described in , or NRP aware TE
(NRP-TE) described in .NRP data plane: Dataplane identifier carried in a data
packet, which is used to mark the link resources and behaviors
allocated to the NRP.NRP steering policy: Policies for steering slice traffic to
the NRP.NRP modification or updates: Modifications or additions to
existing NRP-allocated resources, e.g. some congested links need to
be expanded.NRP monitoring: NRP-allocated resources, including NRP-specific
link or node SID, link bandwidth usage, link delay, and packet loss
status, etc.An NRP is a subset, or all, of resources allocated from a physical
network or logical network. Depending on the SLO and SLE requirements of
the slicing service and also the available resources of the operator's
network, there are several options of creating an NRP. One option is
that each physical link is allocated to only one specific NRP, and
different NRPs do not share any physical link. One more typical option
is that multiple NRPs share the same physical links, and each NRP is
built with virtual links with a certain subset of the bandwidth
available on the physical links to provide network resource
isolation.In addition to specifying resource allocation from the underlay
network, an NRP also needs to have associated control plane and
forwarding plane technologies, which can provide specific routing and
forwarding so that the traffic received from NRP edge nodes that is
characterized to match the NRP traffic classification rule is
constrained to the NRP exclusive topology and resource allocation. The
NRP allows network operators to manage the resources of IETF Network
Slices which are used to provide network slice service traffic with
specific SLOs and SLEs.As defined in , the
draft discusses NRP control plane and data plane requirements in
different provisioning scenarios, and describes that the NRP control
plane is used to exchange network resource attributes and associated
logical topology information between nodes of the NRP so that
NRP-specific routing and forwarding tables could be generated. For the
NRP control plane, distributed control plane mechanism, such as
Multi-topology, Flex-Algo or centralized SDN or hybrid combination could
be defined. To help with forwarding entries, several data-plane
encapsulation options are also discussed to carry NRP information in the
NRP traffic packets. The example NRP data plane identifier could be the
IPv6 addresses or the MPLS forwarding labels or dedicated NRP data-plane
identifiers.An example of NRP instances and a physical network is illustrated in
. In the example, each NRP instance has a customized
network topology comprised of a set of links and nodes in the physical
network. In control plane, each NRP could be associated with a
multi-topology or a Flex-Algo. And it also has its own forwarding plane
resources and identifiers which provide NRP-specific packet
forwarding. also describes the
management of the NRP. After an NRP created, the NRP may need to be
refined and modified as the network status and slice services change,
and could be extended if necessary to meet the customers' demands. In
addition to configuration management, the NRP should also provide
detailed monitoring information about underlying resources to further
provide monitoring for the hosted slice services.One major application of network slices is 5G services. shows the use of the NRP model to realize the IETF
Network Slice for the 5G use case, based on the reference framework
defined in . The
figure shows that the NSC uses the L3VPN network model (L3NM) and the NRP model to map to an IETF Network Slice
service. One possible method is to set the "underlay-transport" of the
L3NM as an NRP instance, which is used to specify the NRP to carry the
VPN traffic. In this way, the NRP-specific resources, together with
NRP control plane and forwarding plane technologies are used to ensure
the SLO and SLE required by the traffic. Similarly, the L2NM can also be used to map an IETF Network Slice
service to an underlying network.In the process of realizing an IETF Network Slice service, the NSC
can use a pre-built NRP instance or dynamically create one as one or a
group of VPNs underlay construct. Compared with current VPN underlay
transport mechanisms, the NRP could provide resource isolation,
topology constraints, and distributed and/ or centralized traffic
engineering (TE). For example, an NRP can use SR policies mechanisms,
such as to optimize the
specific VPN traffic in the NRP topology while providing NRP shortest
path forwarding for other VPN traffic.As defined in , a
network resource partition (NRP) is a collection of resources in the
underlay network. An NRP can have a dedicated topology or can use a
shared topology with other NRPs.Therefore, an NRP is modeled as network topology defined in with augmentations. A new network type "nrp" is
defined. A network topology data instance containing the nrp network
type, indicates an NRP instance. The shows
the relationship between this model and other topology models.The container "nrp" under 'network' of
defines global parameters for an NRP, which defines NRP partition
type, NRP topology generation method, and the specific control plane
and data plane mechanisms of an NRP. And also, the traffic steering
policy of the NRP may include a dynamic color based policies or an
ACL-based static ones.The NRP partition type is used to describe multiple NRP resource
partition methods, for example, no partition, control plane resource
partition, data plane resource partition, or a combination of two
types.As an NRP may consist of the entire or a subset of links in the
underlay network, there are various methods to generate NRP topology,
which include:The NRP with a subset of links in the underlay network, which
has the same topology as the pre-built L3 topology, MT topology,
flexalgo, or TE topology, and also has the same resource
reservation requirements. The topology definition may come
directly from the topology defined by "control plane".For other NRPs that require a dedicated topology,
"nrp-topology-group" is used to configure the selected links from
the base topology. Generally, the base topology refers to the
underlay network topology. An NRP can be configured with one or
more "nrp-topology-group" to create topology resources required by
the NRP. For example, if an NRP needs to reserve the same
bandwidth for a groups of links, the same "group-id" can be
assigned to the links and "bandwidth-reservation" is specified,
such as access network link group, aggregation network link group,
etc. If some inter-domain links, have multiple bandwidth
reservation requirements, they can also be classified into a
group. Then, each link can override the bandwidth reservation of
the group bandwidth reservation.As discussed in , an
NRP could have multiple control plane implementation options. For a
better network scalability, an NRP does not require an independent
distributed control protocol instance or a independent centralized
control plane instance, that is, multiple NRPs can share a same
control plane instance. Thus, an NRP can use a predefined native or
abstract TE topology by referring to a TE network instance or a
predefined control protocol instance by referring to Layer3 network
instance.In addition to global NRP parameters, each NRP instance also
consists of a set of nodes and a set of links, which have different
attributes that represent the allocated resources or the operational
status of the NRP. An NRP could support several data plane resource
partition methods, which are defined by 'link-partition-type'' under
an NRP link, which can further be supported by FlexE or independent
queue techniques.There are multiple modes of NRP operations to be supported as
follows:NRP instantiation: Depending on the slice services types and
also network status, there can be two types of approaches. One
method is to create an NRP instance before the network controller
processes the IETF Network Slice service request. Another one is
that the network controller may start creating an NRP instance
while configuring the IETF Network Slice service request.NRP modification: When the capacity of an existing NPR link is
close to capacity, the bandwidth of the link could be increased.
And when the NRP link or node resources are insufficient, new NRP
links and nodes could be added.NRP Deletion: If the NSC determines that no slice service is
using an NRP, the NSC can delete the NRP instance.NRP Monitoring: The NSC can use the NRP model to track and
monitor NRP resource status and usage.The description of the NRP data nodes are as follows:"nrp-id": Is an identifier that is used to uniquely identify an
NRP instance within the network scope.NRP partition type: Refers to control plane resource partition,
data plane resource partition, or a combination of two types.NRP resources reservation: The nodes and links represent the
network resource allocated for an NRP instance.
'bandwidth-reservation' specifies the bandwidth allocated to an NRP
instance, or is overridden by the configuration of the NRP link.
'link-partition-type' specifies the resource partition types of the
physical interfaces associated with an NRP link.NRP control plane: An NRP can use Multi-Topology Routing (MTR) or
Flex-algo to refer to the IGP instance to generate its own
NRP-specific forwarding tables. Multi-Topology Routing (MTR) is
defined in , , and
or Flex-algo is defined
in .NRP data plane: Defines the data plane mechanism and the NRP
identifier of the network domain managed by the network controller.
The data plane mechanism could be based on MPLS or IPv6 forwarding.
The container "data plane" is used to specify the NRP data plane
encapsulation types and values that are used to identify
NRP-specific network resources. The NRP data plane identifier is
defined, e.g., in and.NRP steering policy: The leaf-list "color-id" is used for dynamic
traffic steering based on SR policy of an NRP and The leaf-list
"acl-ref" is used for common traffic steering.NRP topology group: The list "nrp-topology-group" is used to
explicitly select subset of links of a underlay topology.<CODE BEGINS> file "ietf-nrp@2022-09-26.yang"<CODE ENDS>The YANG model defined in this document is designed to be accessed
via network management protocols such as NETCONF or RESTCONF . The lowest
NETCONF layer is the secure transport layer, and the
mandatory-to-implement secure transport is Secure Shell (SSH) . The lowest RESTCONF layer is HTTPS, and the
mandatory-to-implement secure transport is TLS .The NETCONF access control model 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.There are a number of data nodes defined in this YANG model that are
writable/creatable/deletable (i.e., config true, which is the default).
These data nodes may be considered sensitive or vulnerable in some
network environments. Write operations (e.g., edit-config) to these data
nodes without proper protection can have a negative effect on network
operations.nrp-link: A malicious client could attempt to remove a link from a
topology, add a new link. In each case, the structure of the topology
would be sabotaged, and this scenario could, for example, result in an
NRP topology that is less than optimal.The entries in the nodes above include the whole network
configurations corresponding with the NRP, and indirectly create or
modify the PE or P device configurations. Unexpected changes to these
entries could lead to service disruption and/or network misbehavior.This document registers a URI in the IETF XML registry . Following the format in ,
the following registration is requested to be made:This document requests to register a YANG module in the YANG Module
Names registry .This section contains an example of an instance data tree in JSON
encoding . The example instantiates ietf-nrp for
the topology that is depicted in the following diagram. There are three
nodes, D1, D2, and D3. D1 has three termination points, 1-0-1, 1-2-1,
and 1-3-1. D2 has three termination points as well, 2-1-1, 2-0-1, and
2-3-1. D3 has two termination points, 3-1-1 and 3-2-1. In addition there
are six links, two between each pair of nodes with one going in each
direction.The corresponding NRP instance data tree is depicted below: