TEAS Working Group LM. Contreras
Internet-Draft Telefonica
Intended status: Informational S. Homma
Expires: 27 April 2023 NTT
J.A. Ordonez-Lucena
Telefonica
J. Tantsura
Microsoft
H. Nishihara
NTT-East
October 2022
IETF Network Slice Use Cases and Attributes for the Slice Service
Interface of IETF Network Slice Controllers
draft-ietf-teas-ietf-network-slice-use-cases-01
Abstract
This document analyses the needs of potential customers of network
slices realized with IETF techniques in several use cases, identifies
the functionalities for the IETF Network Slice Service Interface of
an IETF Network Slice Controller to satisfy such requests.
This document is intended to provide motivation and support for work
on YANG models for the IETF Network Slice Service interface.
Status of This Memo
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 4 April 2023.
Copyright Notice
Copyright (c) 2022 IETF Trust and the persons identified as the
document authors. All rights reserved.
Contreras, et al. Expires 27 April 2023 [Page 1]
Internet-Draft IETF Slice Service Interface use cases October 2022
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.
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Conventions used in this document and terminology . . . . . . 4
3. IETF Network Slice Service Interface for IETF Network
Slices . . . . . . . . . . . . . . . . . . . . . . . . . 4
4. IETF Network Slice Use Cases . . . . . . . . . . . . . . . . 5
4.1. 5G Services . . . . . . . . . . . . . . . . . . . . . . . 5
4.1.1. 3GPP network slice . . . . . . . . . . . . . . . . . 6
4.1.1.1. Topology of the TN-NSS . . . . . . . . . . . . . 7
4.1.1.2. Traffic segregation and mapping to S-NSSAI
list . . . . . . . . . . . . . . . . . . . . . . . 7
4.1.1.3. Reachability information . . . . . . . . . . . . 10
4.1.1.4. QoS profiling . . . . . . . . . . . . . . . . . . 10
4.1.2. Private 5G networks . . . . . . . . . . . . . . . . . 11
4.1.2.1. Structure Patterns of Private 5G system . . . . . 11
4.1.2.2. Use Cases Assumed in Private 5G . . . . . . . . . 11
4.1.2.3. Attributes Required in Private 5G . . . . . . . . 12
4.1.3. Generic network Slice Template . . . . . . . . . . . 12
4.1.4. Categorization of GST attributes . . . . . . . . . . 13
4.1.4.1. Attributes with direct impact on the IETF network
slice definition . . . . . . . . . . . . . . . . . 14
4.1.4.2. Attributes with indirect impact on the IETF network
slice definition . . . . . . . . . . . . . . . . . 14
4.1.4.3. Attributes with no impact on the IETF network slice
definition . . . . . . . . . . . . . . . . . . . . 15
4.1.5. Provisioning procedures . . . . . . . . . . . . . . . 16
4.2. NFV-based services . . . . . . . . . . . . . . . . . . . 16
4.2.1. Connectivity attributes . . . . . . . . . . . . . . . 16
4.2.2. Provisioning procedures . . . . . . . . . . . . . . . 17
4.3. Network sharing . . . . . . . . . . . . . . . . . . . . . 18
4.3.1. Connectivity attributes . . . . . . . . . . . . . . . 18
4.3.2. Provisioning procedures . . . . . . . . . . . . . . . 19
4.4. SD-WAN . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.4.1. SD-WAN Structure . . . . . . . . . . . . . . . . . . 19
4.4.2. Connectivity Attributes . . . . . . . . . . . . . . . 21
4.4.3. SD-WAN Endpoint Attributes . . . . . . . . . . . . . 22
4.4.4. SD-WAN UNI Attributes . . . . . . . . . . . . . . . . 23
4.5. Radio functional splits . . . . . . . . . . . . . . . . . 23
Contreras, et al. Expires 27 April 2023 [Page 2]
Internet-Draft IETF Slice Service Interface use cases October 2022
4.5.1. Attributes and procedures . . . . . . . . . . . . . . 24
4.6. Data center interconnection . . . . . . . . . . . . . . . 24
4.6.1. Connectivity attributes . . . . . . . . . . . . . . . 25
4.6.2. Provisioning procedures . . . . . . . . . . . . . . . 25
5. Summary of attributes and procedures . . . . . . . . . . . . 25
5.1. Summary of SLOs . . . . . . . . . . . . . . . . . . . . . 25
5.2. Summary of SLEs . . . . . . . . . . . . . . . . . . . . . 26
5.3. Summary of procedures . . . . . . . . . . . . . . . . . . 26
6. Gaps on related IETF Network Slice efforts . . . . . . . . . 26
7. Security Considerations . . . . . . . . . . . . . . . . . . . 26
8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 27
9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 27
10. References . . . . . . . . . . . . . . . . . . . . . . . . . 27
10.1. Normative References . . . . . . . . . . . . . . . . . . 27
10.2. Informative References . . . . . . . . . . . . . . . . . 27
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 28
1. Introduction
A number of new technologies, such as 5G, NFV and SDN are not only
evolving the network from a pure technological perspective but also
are changing the concept in which new services are offered to the
customers [I-D.homma-slice-provision-models] by introducing the
concept of network slicing.
The transport network is an essential component in the end-to-end
delivery of services and, consequently, it is necessary to understand
what could be the way in which the transport network is consumed as a
slice. For a definition of IETF network slice refer to
[I-D.ietf-teas-ietf-network-slices].
In this document it is assumed that there exists a (logically)
centralized component in the transport network, namely IETF Network
Slice Controller (NSC) with the responsibilities on the control and
management of the IETF network slices invoked for a given service, as
requested by IETF network slice customers.
This document analyses different use cases deriving the needs of
potential IETF network slice customers in order to identify the
functionality required on the IETF Network Slice Service Interface of
the NSC to be exposed towards such IETF network slice customers.
With that purpose, this document is intended to provide motivation
and support for work on YANG models for the IETF Network Slice
Service interface (such as e.g.
[I-D.ietf-teas-ietf-network-slice-nbi-yang]). As such, it might not
be necessary to advance it to publication as an RFC.
Contreras, et al. Expires 27 April 2023 [Page 3]
Internet-Draft IETF Slice Service Interface use cases October 2022
Solutions to construct the requested IETF network slices are out of
scope of this document.
2. Conventions used in this document and terminology
The keywords "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
BCP14, [RFC2119], [RFC8174] when, and only when, they appear in all
capitals, as shown here.
Regarding terminology, the term "transport network" in the context of
this document refers in broad sense to WAN, MBH, IP backbone and
other network segments implemented by IETF technologies.
3. IETF Network Slice Service Interface for IETF Network Slices
In a general manner, the transport network supports different kinds
of services. These services consume capabilities provided by the
transport network for deploying end-to-end services, interconnecting
network functions or applications spread across the network and
providing connectivity toward the final users of these services.
Under the slicing approach, a IETF network slice customer requests to
a IETF network slice controller a slice with certain characteristics
and parametrization. Such request it is assumed here to be done
through a IETF Network Slice Service Interface exposed by the NSC to
the customer, as reflected in Figure 1.
+--------------------+
| |
| IETF Network |
| Slice Customer |
| |
+--------------------+
A
|
| IETF Network
| Slice Service
| Interface
|
V
+--------------------+
| |
| IETF Network |
| Slice Controller |
| |
+--------------------+
Contreras, et al. Expires 27 April 2023 [Page 4]
Internet-Draft IETF Slice Service Interface use cases October 2022
Figure 1: IETF Network Slice Service Interface concept
The functionality supported by the IETF Network Slice Service
Interface depends on the service objectives that the slice customer
requires. It is then important to understand the needs of the slice
customers as well as the way of expressing them.
4. IETF Network Slice Use Cases
Different use cases for slice customers can be identified, as
described in the following sections.
4.1. 5G Services
5G services natively rely on the concept of network slicing. 5G is
expected to allow vertical customers to request slices in such a
manner that the allocated resources and capabilities in the network
appear as dedicated for them.
In network slicing scenarios, a vertical customer requests a network
operator to allocate a network slice instance (NSI) satisfying a
particular set of service requirements. The content/format of these
requirements are highly dependent on the networking expertise and use
cases of the customer under consideration. To deal with this
heterogeneity, it is fundamental for the network operator to define a
a unified ability to interpret service requirements from different
vertical customers, and to represent them in a common language, with
the purposes of facilitating their translation/mapping into specific
slicing-aware network configuration actions. In this regard, model-
based network slice descriptors built on the principles of
reproducibility, reusability and customizability can be defined for
this end.
Contreras, et al. Expires 27 April 2023 [Page 5]
Internet-Draft IETF Slice Service Interface use cases October 2022
As a starting point for such a definition, GSMA developed the idea of
having a universal blueprint that, being offered by network
operators, can be used by any vertical customer to order the
deployment of an NSI based on a specific set of service requirements.
The result of this work has been the definition of a baseline network
slice descriptor called Generic network Slice Template (GST). The
GST contains multiple attributes that can be used to characterize a
network slice. A Network Slice Type (NEST) describes the
characteristics of a network slice by means of filling GST attributes
with values based on specific service requirements. Basically, a
NEST is a filled-in version of a GST. Different NESTs allow
describing different types of network slices. For slices based on
standardized service types, e.g. eMBB, uRLLC and mIoT, the network
operator may have a set of readymade, standardized NESTs (S-NESTs).
For slices based on specific industry use cases, the network operator
can define additional NESTs.
Service requirements from a given vertical customer are mapped to a
NEST, which provides a self-contained description of the network
slice to be provisioned for that vertical customer. According to
this reasoning, the NEST can be used by the network operator as input
to the NSI preparation phase, which is defined in [TS28.530]. 3GPP is
working on the translation of the GST/NEST attributes into NSI
related requirements, which are defined in the "ServiceProfile" data
type from the Network Slice Information Object Class (IOC) in
[TS28.541]. These requirements are used by the 3GPP Management
System to allocate the NSI across all network domains, including
transport network. The IETF network slice defines the part of that
NSI that is deployed across the transport network.
Despite the translation is an on-going work in 3GPP it seems
convenient to start looking at the GST attributes to understand what
kind of parameters could be required for the IETF Network Slice
Service Interface.
4.1.1. 3GPP network slice
A 3GPP network slice represents a logical network that provides
specific capabilities and network characteristics, supporting the
service requirements of one or more network slice customers. The
service requirements of each network slice customer are captured into
a separate "ServiceProfile" artifact within the network slice class
(see Network Slicing NRM fragment in TS 28.541).
A 3GPP network slice spans from 5GNR access nodes to the UPF that
terminates the PDU session, i.e. PSA UPF. In this in-slice data
path, there are TN segments (e.g. backhaul) that are out of scope of
3GPP management domain. For the provisioning and operation of these
Contreras, et al. Expires 27 April 2023 [Page 6]
Internet-Draft IETF Slice Service Interface use cases October 2022
TN segments, usually referred to as transport Network Slice Subnets
(TN-NSS), the 3GPP management system relies on an external TN
management system, which hosts (among other components) the IETF NSC.
To proceed with this delegation, the 3GPP management system needs to
make available to the TN management system the information described
in the following sub-sections.
4.1.1.1. Topology of the TN-NSS
The TN management system needs to know the transport termination/end
points to determine the transport resources, either physical or
virtual nodes. 3GPP management system systems need to provide the
transport endpoints of 3GPP managed functions that are part of the
RAN-NSS (e.g., gNB-CU-UP, gNB-CU-CP) and CN-NSS (e.g., UPF, AMF), and
if applicable further information such as the next-hop router IP
address configured in a RAN-NSS or CN-NSS. The TN management system
should be able to correlate this with the transport network topology
and derive the site or border routers connecting to 3GPP managed
functions.
4.1.1.2. Traffic segregation and mapping to S-NSSAI list
As network functions can be shared by many network slices, it will be
necessary to segregate the traffic belonging to specific slices on
transport interfaces.
One option for traffic segregation is to assign application endpoints
to specific sets of S-NSSAI values. The transport network can map
packets to connectivity services based on local remote or remote
endpoints, provided that the allocation of S-NSSAI to endpoints is
known and exposed, and provided that the application endpoints are
visible on the transport layer. The application endpoints visible in
a RAN-NSS and CN-NSS are already mapped to a specific set of S-NSSAI.
Figure 2 illustrates an example of this solution, whereby a 3GPP
network slice with S-NSSAI=1 is mapped to specific application
endpoints (e.g., N3 tunnel endpoint 1) by the access network node.
In this example, the TN management system decides to map application
endpoints 1 and 2 to the same transport connectivity service A. This
mapping is implemented by the site router connecting to the access
network node. On the core network slice, a similar mapping is done
by the border router. Demultiplexing the packet streams belonging to
different transport interfaces is based on regular routing and
reachability of endpoint IP addresses.
Contreras, et al. Expires 27 April 2023 [Page 7]
Internet-Draft IETF Slice Service Interface use cases October 2022
Transport Interface Transport Interface
(App_EP "x" <-> CSR) (BR <-> App_EP "x")
+--------------+ +-------+ +-------+ +--------------+
+---|----+ +----|----+ | | | | +---|-----+ +---|----+
|NSS-AN 1|----| App_EP 1|<-->|-- | | --|<-->|App_EP 1 |----|NSS-CN 1|
+---|----+ +----|----+ | \ +-|--Transport-----+ / | +---|-----+ +---|----+
| | | --| Connectivity |-- | | |
+---|----+ +----|----+ | / +-|-----"A"--------+ \ | +---|-----+ +---|----+
|NSS-AN 2|----| App_EP 2|<-->|-- | | --|<-->|App_EP 2 |----|NSS-CN 2|
+---|----+ +----|----+ | | | | +---|-----+ +---|----+
| | | | | | | |
+---|----+ | | | | | | +---|----+
|NSS-AN 3|- | | | | | | |NSS-CN 3|
+---|----+ \ +----|----+ | +----Transport-----+ | +---|-----+ +---|----+
| --| App_EP 3|<-->|-----| Connectivity |-----|<-->|App_EP 3 |-- |
+---|----+ / +----|----+ | +-------"B"--------+ | +---|-----+ \ +---|----+
|NSS-AN 4|- | | | | | | -|NSS-CN 4|
+---|----+ | | | | | | +---|----+
+--------------+ +-------+ +-------+ +--------------+
Access network node(s) Cell Site Router (CSR) Border Router (BR) Core network node(s)
S-NSSAI=1: {NSS-AN 1, App_EP 1, Transport Connectivity A, App_EP 1, NSS-CN 1}
S-NSSAI=2: {NSS-AN 2, App_EP 2, Transport Connectivity A, App_EP 2, NSS-CN 2}
S-NSSAI=4: {NSS-AN 4, App_EP 3, Transport Connectivity B, App_EP 3, NSS-CN 3}
Figure 2: Mapping of S-NSSAI to specific application endpoints
Despite the simplicity of the above-referred approach, notice that it
is not a universal solution as the application endpoint addresses are
not always visible to the TN, for example when they are encrypted by
IPSec tunnels. In such a case, the application endpoints are not
visible to the site router, and thus cannot be used for transport
connectivity mappings. To deal with these situations, an alternative
solution is to use the concept of logical transport interfaces. A
logical transport interface is a virtual interface separate from
application endpoints; it can be for example a specific IP address /
VLAN combination that corresponds to an IPSec termination point, an
identifier (e.g., MPLS label, segment ID) that the TN recognizes, or
it can be just a logical interface defined on top of top a physical
transport interface. As long as the interface identity can derived
from packet headers, the TN nodes can perform the mapping to
transport connectivity services. In this regard, it is useful to
indicate to the TN which traffic types are carried over an interface
(e.g., N3 user plane packets, N2 control plane packets, etc.).
Contreras, et al. Expires 27 April 2023 [Page 8]
Internet-Draft IETF Slice Service Interface use cases October 2022
Figure 3 illustrates an example on the use of this solution. As
seen, logical transport need to be exposed from 3GPP management
system to TN management system, so that the latter can create
transport network topology and determine the TN resources to support
the 3GPP slice.
Logical transport interface, Logical transport interface,
exposed by RAN NSSMF exposed by CN NSSMF
+--------+ +-------+ +-------+ +-------+
| | | | | | | |
+---|----+ +----Logical-----+ | | +----Logical----+ +---|----+
|NSS-AN 1|--| Transport |- | | -| Transport |--|NSS-CN 1|
+---|----+ +--Interface 1---+ \ +---Transport----+ / +--Interface 1--+ +---|----+
| | | --| Connectivity |-- | | |
+---|----+ +----Logical-----+ / +-------"A"------+ \ +----Logical----+ +---|----+
|NSS-AN 2|--| Transport |- | | -| Transport |--|NSS-CN 2|
+---|----+ +--Interface 2---+ | | +--Interface 1--+ +---|----+
| | | | | | | |
+---|----+ | | | | | | +---|----+
|NSS-AN 3|- | | | | | | -|NSS-CN 3|
+---|----+ \+----Logical-----+ +---Transport---+ +----Logical-----+/ +---|----+
| | Transport |----| Connectivity |----| Transport | |
+---|----+ /+--Interface 2---+ +-------"B"-----+ +--Interface 1---+\ +---|----+
|NSS-AN 4|- | | | | | | -|NSS-CN 4|
+---|----+ | | | | | | +---|----+
| | | | | | | |
+--------+ +-------+ +-------+ +-------+
Access network Cell Site Border Router Core network
node(s) Router (CSR) (BR) node(s)
S-NSSAI=1: {NSS-AN 1, Logical Transport Interface 1, Transport Connectivity A,
Logical Transport Interface 1, NSS-CN 1}
S-NSSAI=2: {NSS-AN 2, Logical Transport Interface 2, Transport Connectivity A,
Logical Transport Interface 2, NSS-CN 2}
S-NSSAI=4: {NSS-AN 4, Logical Transport Interface 3, Transport Connectivity B,
Logical Transport Interface 3, NSS-CN 4}
Figure 3: Logical Transport Interfaces
For traffic segregation, though solutions might be valid, 3GPP
prefers the second solution: on the use of concept of transport
logical interface. The reason is that it does not impose 1:1 mapping
between application endpoint and transport interface (allowing for
better redundancy) and that it always works, no matter if encryption.
To support this solution, the 3GPP has recently extended the Network
Slice NRM fragment, including a new Information Object Class called
Contreras, et al. Expires 27 April 2023 [Page 9]
Internet-Draft IETF Slice Service Interface use cases October 2022
EP_Transport. This class provides a complete characterization of the
logical transport interface, including transport level information
(i.e., IP address, reachability information, QoS profile) and the set
of application endpoints aggregated to this interface. For further
information on reachability information and QoS profile, see next
subsections. For further details on fields of EP_Transport, see
Network Slice NRM fragment in TS 28.541.
4.1.1.3. Reachability information
Each physical or logical transport interface will carry the traffic
associated with some 3GPP application endpoints that may be using IP
addresses separate from the transport interface. These IP addresses
must be reachable within the TN-NSS, and hence they need to be
advertised to populate forwarding tables. A 3GPP network function
can advertise such reachability information by running a dynamic
routing protocol towards the next hop router. If that is not
possible, it can create association between the reachability data
with the logical transport interface and expose it towards the 3GPP
and TN management system. This information can be derived from the
IP addresses available for application and transport endpoints.
4.1.1.4. QoS profiling
Each TN-NSS may be associated a "TNSliceSubnetProfile", which hosts
the SLO requirements (e.g., guaranteed throughput, bounded latency,
maximum jitter) that the TN-NSS must support. "TNSliceSubnetProfile"
is a 3GPP artifact that result from the decomposition of e2e service
requirements ("ServiceProfile" artifact ) into domain-specific
service requirements ("RANSliceSubnetProfile", "CNSliceSubnetProfile"
and "TNSliceSubnetProfile") applicable to RAN-NSS, CN-NSS and TN-NSS
respectively. Unlike "RANSliceSubnetProfile" and
"CNSliceSubnetProfile", there is not agreement yet on the specific
parameters to be captured by the "TNSliceSubnetProfile". Further
work in this regard in the upcoming 3GPP SA5 meetings.
Upon receiving the "TNSliceSubnetProfile" from the 3GPP management
system, the TN management system translates the SLO requirements
therein into a QoS profile, which includes applicability and use of
DSCPs and other QoS related properties onto the TN-NSS realization.
To enable this, each logical interface may have an associated QoS
profile. The QoS profile is just a reference to the detailed profile
parameters which are logically provisioned on both sides of a logical
transport interface.
Contreras, et al. Expires 27 April 2023 [Page 10]
Internet-Draft IETF Slice Service Interface use cases October 2022
4.1.2. Private 5G networks
Private 5G is one of variations of 5G service provision. Private 5G
allows unlicensed as well as licensed companies to establish and
operate 5G networks, with frequency band assigned for private 5G, in
their own companies.
Private 5G can be customized flexibly rather than public 5G, and thus
it enables us to provide networks specialized for their use cases.
Private 5G is also called non-public 5G, and its deployment scenarios
and service attributes are described in (ref. [TS23.501]).
4.1.2.1. Structure Patterns of Private 5G system
In Private 5G, a Service Provider does not necessarily have its own
resources (e.g., radio bases, transit network and server resources
for 5G CP functions) and can flexibly customize and deploy by
selecting and combining various resources.
Private 5G has several structure patterns:
* Pattern 1: a service provider has all resources including radio
bases, transit networks, and server resources for 5G CP functions.
* Pattern 2: a service provider has radio bases and server resources
for 5G CP functions, and lends transit networks from other network
operators.
* Pattern 3: a service provider has only radio bases and lends
transit networks and server resources for 5G CP functions from
other network operators and data center companies.
In pattern 2 and 3, it is assumed that a service provider uses
network slices provided by other companies.
4.1.2.2. Use Cases Assumed in Private 5G
Private 5G provides a wireless communication environment which has
specific features depending on applications or usage, within limited
areas. From such aspects, within 5G use cases (ref. [TS22.261]),
the following communication types and use cases could be especially
expected to be provided with private 5G.
* High-bandwidth and reliable communication:
- VR streaming
* Low latency and jitter:
Contreras, et al. Expires 27 April 2023 [Page 11]
Internet-Draft IETF Slice Service Interface use cases October 2022
- Smart factory
- Remote automated robot operation (e.g., robot concierge/
assistant, robot waiter, drone)
* High-bandwidth on up-link and low latency and jitter
- Remote surgery
- Uploading of high-definition video
4.1.2.3. Attributes Required in Private 5G
Private 5G has some distinguished requirements to network slice as
below.
* QoS customization:
- assured bandwidth
- assured latency and jitter
- customization of UL/DL rate on throughput (e.g., for video
upstreaming consumes much UL bandwidth)
* Multi-homing (for high reliability, preparing multiple paths
traverse different physical routes)
* Performance monitoring (e.g., for connectivity status and service
availability of devices)
* Traffic flow separation/segregation (e.g., segregation of user
plane and other communications physically and/or logically)
4.1.3. Generic network Slice Template
The structure of the GST is defined in [GSMA]. The template defines
a total of 35 attributes. For each of them, the following
information is provided:
* Attribute definition, which provides a formal definition of what
the attribute represents.
* Attribute parameters, including:
- Value, e.g. integer, float.
- Measurement unit, e.g. milliseconds, Gbps
Contreras, et al. Expires 27 April 2023 [Page 12]
Internet-Draft IETF Slice Service Interface use cases October 2022
- Example, which provides examples of values the parameter can
take in different use cases.
- Tag, which allow describing the type of parameter, according to
its semantics. An attribute can be tagged as a
characterization attribute or a scalability attribute. If it
is characterization attribute, it can be further tagged as a
performance-related attribute, a functionality-related
attribute or an operation-related attribute.
- Exposure, which allow describing how this attribute interact
with the slice customer, either as an API or a KPI.
* Attribute presence, either mandatory, conditional or optional.
Attributes from GST can be used by the network operator (slice
controller) and a vertical customer (slice customer) to agree SLA.
GST attributes are generic in the sense that they can be used to
characterize different types of network slices. Once those
attributes become filled with specific values, it becomes a NEST
which can be ordered by slice customers.
4.1.4. Categorization of GST attributes
Not all the GST attributes as defined in [GSMA] have impact in the
transport network since some of them are specific to either the radio
or the mobile core part.
In the analysis performed in this document, the attributes have been
categorized as:
* Directly impactive attributes, which are those that have direct
impact on the definition of the IETF network slice, i.e.,
attributes that can be directly translated into requirements
required to be satisfied by a IETF network slice.
* Indirectly impactive attributes, which are those that impact in an
indirect manner on the definition of the IETF network slice, i.e.,
attributes that indirectly impose some requirements to a IETF
network slice.
* Non-impactive attributes, that are those which do not have impact
on the IETF network slice at all.
The following sections describe the attributes falling into the three
categories.
Contreras, et al. Expires 27 April 2023 [Page 13]
Internet-Draft IETF Slice Service Interface use cases October 2022
4.1.4.1. Attributes with direct impact on the IETF network slice
definition
The following attributes impose requirements in the IETF network
slice
* Availability
* Deterministic communication
* Downlink throughput per network slice
* Energy efficiency
* Group communication support
* Isolation level
* Maximum supported packet size
* Mission critical support
* Performance monitoring
* Slice quality of service parameters
* Support for non-IP traffic
* Uplink throughput per network slice
* User data access (i.e., tunneling mechanisms)
4.1.4.2. Attributes with indirect impact on the IETF network slice
definition
The following attributes indirectly impose requirements in the IETF
network slice to support the end-to-end service.
* Area of service (i.e., the area where terminals can access a
particular network slice)
* Delay tolerance (i.e., if the service can be delivered when the
system has sufficient resources)
* Downlink (maximum) throughput per UE
* Network functions owned by Network Slice Customer
Contreras, et al. Expires 27 April 2023 [Page 14]
Internet-Draft IETF Slice Service Interface use cases October 2022
* Maximum number of (concurrent) PDU sessions
* Performance prediction (i.e., capability to predict the network
and service status)
* Root cause investigation
* Session and Service Continuity support
* Simultaneous use of the network slice
* Supported device velocity
* UE density
* Uplink (maximum) throughput per UE
* User management openness (i.e., capability to manage users'
network services and corresponding requirements)
* Latency from (last) UPF to Application Server
4.1.4.3. Attributes with no impact on the IETF network slice definition
The following attributes do not impact the IETF network slice.
* Location based message delivery (not related to the geographical
spread of the network slice itself but with the localized
distribution of information)
* MMTel support, i.e. support of and Multimedia Telephony Service
(MMTel)as well as IP Multimedia Subsystem (IMS) support.
* NB-IoT Support, i.e., support of NB-IoT in the RAN in the network
slice.
* Maximum number of (simultaneous) UEs
* Positioning support
* Radio spectrum
* Synchronicity (among devices)
* V2X communication mode
* Network Slice Specific Authentication and Authorization (NSSAA)
Contreras, et al. Expires 27 April 2023 [Page 15]
Internet-Draft IETF Slice Service Interface use cases October 2022
4.1.5. Provisioning procedures
3GPP identifies in [TS28.541] a number of procedures for the
provisioning of a network slice in general. It can be assumed that
similar procedures may also apply to a transport slice, facilitating
a consistent management and control of end-to-end slices.
The envisioned procedures are the following:
* Slice instance allocation: this procedure permits to create a new
slice instance (or reuse an existing one).
* Slice instance de-allocation: this procedure decommissions a
previously instantiated slice.
* Slice instance modification: this procedure permits the change in
the characteristics of an existing slice instance.
* Get slice instance status: this procedure helps to retrieve run-
time information on the status of a deployed slice instance.
* Retrieval of slice capabilities: this procedure assists on getting
information about the capabilities (e.g. maximum latency
supported).
All these procedures fit in the operation of transport network
slices.
4.2. NFV-based services
NFV technology allows the flexible and dynamic instantiation of
virtualized network functions (and their composition into network
services) on top of a distributed, cloud-enabled compute
infrastructure. This infrastructure can span across different points
of presence in a carrier network. By leveraging on transport network
slicing, connectivity services established across geographically
remote points of presence can be enriched by providing additional QoS
guarantees with respect present state-of-the-art mechanisms, as
conventional L2/L3 VPNs.
4.2.1. Connectivity attributes
The connectivity services are expressed through a number of
attributes as listed:
* Incoming and outgoing bandwidth: bandwidth required for the
connectivity services (in Mbps).
Contreras, et al. Expires 27 April 2023 [Page 16]
Internet-Draft IETF Slice Service Interface use cases October 2022
* Qos metrics: set of metrics (e.g., cost, latency and delay
variation) applicable to a specific connectivity service
* Directionality: indication if the traffic is unidirectional or
bidirectional.
* MTU: value of the largest PDU to be transmitted in the
connectivity service.
* Protection scheme: indication of the kind of protection to be
performed (e.g., 1;1, 1+1, etc.)
* Connectivity mode: indication of the service is point-to-point of
point-to-multipoint
All those attributes will assist on the characterization of the
connectivity slice to be deployed, and thus, are relevant for the
definition of a IETF network slice supporting such connectivity.
4.2.2. Provisioning procedures
ETSI NFV defines the role of WAN Infrastructure Manager (WIM) as the
component in charge of managing and controlling the connectivity
external to the PoPs. In [IFA032] a number of interfaces are
identified to be exposed by the WIM for supporting the multi-site
connectivity, thus representing the capabilities expected for a
transport network slice, as well, in case of satisfying such
connectivity needs by means of the slice concept.
The interfaces considered are the following:
* Multi-Site Connectivity Service (MSCS) Management: this interface
permits the creation, termination, update and query of MSCSs,
including reservation. It also enables subscription for
notifications and information retrieval associated to the
connectivity service.
* Capacity Management: this interface allows querying about the
capacity (e.g. bandwidth), topology, and network edge points of
the connectivity service, as well as about information of consumed
and available capacity on the underlying network resources.
* Fault Management: this interface serves for the provision of
alarms related to the MSCSs.
* Performance Management: this interface assists on the retrieval of
performance information (measurement results collection and
notifications) related to MSCSs.
Contreras, et al. Expires 27 April 2023 [Page 17]
Internet-Draft IETF Slice Service Interface use cases October 2022
4.3. Network sharing
Network sharing is one of the means network operators exploit for
increasing efficiencies. There are different scenarios of network
sharing, being especially popular in the deployment of mobile
networks, typically referred to as Radio Access Network (RAN)
sharing. From an operational perspective, in RAN sharing we have two
roles: master operator, being the actor (e.g. infrastructure
provider, network operator) to which the deployment and daily
operation of shared RAN elements are entrusted to; and the
participant operators, who are the mobile operators who share the RAN
facilities provided by the master operator. Note that in this
context the master and participant operator can be seen as provider
and customer, respectively.
While there exist different modes of RAN sharing [TS23.251],
including passive RAN sharing (infrastructure site sharing) and
active RAN sharing (e.g. Multi-Operator Core Networks or MOCN), most
of the cases require the establishment of separated connections in
order to separate the traffic per participant operator. Such
connections typically extend from the cell site to some pre-defined
and agreed interconnection points, from which the traffic is routed
and delivered to individual participant operators.
The above-referred connections can have specific attributes. Aspects
like guaranteed bandwidth (in line with the expected load from the
aggregated cells), redundancy, bounded latency (per kind of traffic),
or secure delivery of the information should be considered.
The master operator is the one in charge of provisioning the
connections and collecting management data (e.g. performance
measurements, telemetry, fault alarms, trace data) for individual
participant operators. The use of network slicing could make the
network sharing approach more flexible by allowing the other
operators control and manage the established connections [MEF].
The implications of the RAN sharing scenario here described can be
extended to either fixed networks or even to mobile networks
leveraging on radio functional split (i.e., including fronthaul and
midhaul network segments).
4.3.1. Connectivity attributes
The connections for RAN sharing typically consider attributes like:
* Maximum and Guaranteed Bit Rate (MBR and GBR respectively).
* Bounded latency (e.g., for user plane, control plane, etc)
Contreras, et al. Expires 27 April 2023 [Page 18]
Internet-Draft IETF Slice Service Interface use cases October 2022
* Packet loss rate.
* IP addressing (consistent among the operators sharing the
infrastructure).
* L2/L3 reachability.
* Recovery time (on the event of failures).
* Secure connection (e.g., encryption support).
4.3.2. Provisioning procedures
The expected provisioning procedures are:
* Connection provisioning between site and interconnection point.
Those connections could evolve in time in terms of capacity
depending on the capacity growth of each particular site.
* Collection of management data, including performance measurements,
fault alarms and trace data.
4.4. SD-WAN
SD-WAN is a solution to provide a virtual overlay network for
connecting between customer's sites, (virtual) private cloud, or
public cloud/Internet. SD-WAN operates over one or more underlay
networks, and enables to offer more differentiated service delivery
capabilities. SD-WAN can be esteemed as a type of network slices or
can be established over underlay networks provided as network slices.
The definitions, specification, service attributes, and framework of
SD-WAN is defined in Metro Ethernet Forum ([MEF-70]).
SD-WAN forwards traffic based on application flows, and the policies
include rules and constraints on the forwarding of the application
flows. In SD-WAN, it may be required from the customer to adjust the
behaviors based on its needs in near real time. The service provider
is required to monitor the performance of the service and modify the
forwarding policies based on the real-time telemetry from the
underlying network components.
4.4.1. SD-WAN Structure
SD-WAN has three logical constructs:
* SD-WAN virtual connection
* SD-WAN virtual connection endpoint
Contreras, et al. Expires 27 April 2023 [Page 19]
Internet-Draft IETF Slice Service Interface use cases October 2022
* SD-WAN UNI
Several additional components may be visible to the customer. These
include:
* Customer network
* Service provider network
* Underlay connectivity
* Tunnel virtual connection
The following figure shows the overview of SD-WAN structure. In this
case, the customer sites are connected with underlay connectivity#1
and they are also connected to remote private cloud with underlay
connectivity#2. An SD-WAN endpoint is usually located in each
customer network site as a CPE or a customer edge, and it allocates
application flow to appropriate underlay connectivity.
,----.
| \
,-/ Private`--.
| cloud |
`---+---+-----/
- - - - |EP | - - - - -
| +---+ |
| # |
/---------\| # |/----------\
| Customer +--+=========#===========+--+ Customer |
| Network |EP|. . . . . . . . . . .|EP| Network |
| site A +--+ Service Provider +--+ site B |
\--------/| Network |\--------/
| - - - - - - - - - - - - - |
| |
SD-WAN UNI SD-WAN UNI
* Legend
. . . : Underlay connectivity#1
===== : Underlay Connectivity#2
EP : SD-WAN Endpoint
Figure 4: Overview of SD-WAN Structure
Contreras, et al. Expires 27 April 2023 [Page 20]
Internet-Draft IETF Slice Service Interface use cases October 2022
SD-WAN may be provided as a network slice, or it is realized on
several network slices provided as underlay connectivities. In the
former case, a network slice PE will be mapped to CE in SD-WAN. In
the later case, PEs of the provider of underlay connectivities will
behave as network slice PEs.
4.4.2. Connectivity Attributes
SD-WAN defined in MEF-70 has several attributes on its connectivity
as below:
o SD-WAN Identifier: the value is a string that is used by the
customer and service provider to uniquely identify an SD-WAN
connectivity.
o Endpoint list: the value is a list contains endpoint identifiers
and their connected endpoints.
o Service Uptime Objective: the value is the proportion of time that
the connectivity service is working during a given time period.
o Reserved Prefixes: the values are IP prefixes reserved by the
service provider for use for SD-WAN within its own network or for
distribution to the customer via DHCP or SLAAC.
o List for Policies: the value is a list of policies applied to
application flows and application flow groups at endpoints. An
SD-WAN policy list contains policy name and list of policy
criteria. Support of the criteria listed below would be required:
* Encryption: indicates whether or not the application flow
requires encryption
* Public-Private: indicates whether the application flow can
traverse public or private underlay connectivity services (or
both).
* Internet-Breakout: indicates whether the application flow
should be forwarded to an Internet destination.
* Billing-Method: indicate the application flow can be sent over
an underlay connectivity service that has usage-based or flat-
rate billing.
* Backup: indicates whether this application flow can use a TVC
designated as aEURbackupaEUR.
* Bandwidth: specifies a rate limit on the application flow.
Contreras, et al. Expires 27 April 2023 [Page 21]
Internet-Draft IETF Slice Service Interface use cases October 2022
o List of Application Flow Groups: the value is a list of
application flow groups that application flows can be members of.
An application flow group list contains application flow group
name and application flow group policy.
o List of Application Flows: the value is a list of the application
flows that are recognized by the SD-WAN. An application flow list
contains application flow name, list of application flow criteria,
and application flow group name. The criteria is listed below:
* Ethertype
* C-VLAN ID list
* IPv4 source address
* IPv4 destination address
* IPv4 source or destination address
* IPv4 protocol list
* IPv6 source address
* IPv6 destination address
* IPv6 source or destination address
* IPv6 next header list
* TCP/UDP source port list
* TCP/UDP destination port list
* Application identifier
* any
4.4.3. SD-WAN Endpoint Attributes
SD-WAN contains some endpoints as boundary nodes between underlay
connections and customers sites. [MEF-70] defines some attributes
for SD-WAN endpoints as below:
o Endpoint Identifier: the value is for identification of SD-WAN
endpoint for management purposes.
o Endpoint UNI: the value is for identification of the UNI that the
Contreras, et al. Expires 27 April 2023 [Page 22]
Internet-Draft IETF Slice Service Interface use cases October 2022
endpoint is associated with.
o Endpoint policy map: the value is for mapping policies to
application flows and application flow groups.
4.4.4. SD-WAN UNI Attributes
SD-WAN UNI is a reference point that represents the demarcation
between the responsibility of the customer and the responsibility of
the provider. Some attributes for UNI is defined in [MEF-70] as
below:
o SD-WAN UNI Identifier: the value is for identification of the UNI
for management purposes.
o SD-WAN UNI L2 Interface: the value describes the underlay L2
interface for the UNI.
o SD-WAN UNI Maximum L2 Frame Size: the value specifies the maximum
length L2 frame that is accepted by the provider.
o SD-WAN UNI IPv4 connection addressing: the value describes IPv4
connection address mechanisms (e.g., Static or DHCP).
o SD-WAN UNI IPv6 connection addressing: the value describes IPv6
connection address mechanisms (e.g., DHCP, SLAAC, Static or Link-
Local-only).
4.5. Radio functional splits
The disaggregation of the software stack in radio base stations
allows the centralization of some of the radio processing functions.
O-RAN is promoting the interoperability of implementations of radio
functional splits, defining an architecture where three main entities
can be considered: the Radio Unit (RU), with some basic processing,
the Distributed Unit (DU) with the rest of real-time processing
capabilities, and the Centralized Unit (CU) with the non-real-time
processing of the software stack. The network segment between RU and
DU is known as fronthaul (FH), while the segment between DU and CU is
referred as midhaul (MH). Figure 5 shows this situation.
Contreras, et al. Expires 27 April 2023 [Page 23]
Internet-Draft IETF Slice Service Interface use cases October 2022
.........................................
: Radio functional split :
+-------+ : :
| radio | :+----+ Fronthaul +----+ Midhaul +----+ : Backhaul +-----+
| base | <=> :| RU |<=========>| DU |<=======>| CU | :<========>| UPF |
|station| :+----+ +----+ +----+ : +-----+
+-------+ : :
: :
:.......................................:
Figure 5: Logical Transport Interfaces
The fronthaul leverages on eCPRI protocol which can be transported
directly on Ethernet frames or encapsulated in IP/UDP (for the user
plane). The midhaul can be transported in a similar way as the
backhaul.
With current specifications, individual service flows being carried
by FH cannot be distinguished, so no possibility of differentiating
connectivity slices at that point. Similar thing happens for MH.
The only possible differentiation per flow can happen in downstream
direction from CU to DU, but this basically can only help for
policing traffic at that point (i.e., slice is yet the same).
Advanced scenarios such as RU sharing could allow traffic
differentiation per mobile operator based on e.g. vlans, being each
of those vlans mapped to a different slice.
4.5.1. Attributes and procedures
The attributes of IETF network slices for the conveniently supported
the radio functional split are based on main characteristics of FH/
MH: Latency, BW, and packet loss, as specified in [O-RAN].
Geographical location could have an impact due to latency
restrictions for FH.
Regarding slice management procedures, it can be assumed a similar
lifecycle as in 3GPP slices.
4.6. Data center interconnection
The distribution of computing facilities in the form of
interconnected data centers is changing the traffic profiles observed
on operators' networks. For some applications is necessary to
provide guarantees over the WAN connection in the form of a network
slice.
Contreras, et al. Expires 27 April 2023 [Page 24]
Internet-Draft IETF Slice Service Interface use cases October 2022
In many large data centers, Remote Direct Memory Access (RDMA) is
becoming the prevalent type traffic. The growth of this traffic is
driven by the trend of storage disaggregation and the offload
capabilities enabled by RDMA. RDMA can be carried on top of IP/UDP,
in which is known as RDMA over Converged Ethernet (RoCE).
Given the overall growth in DC cluster sizes per location and related
limitations e.g. in terms of power consumption and cooling, the RDMA
traffic can require to be delivered across distributed data centers,
thus impacting the Data Center Interconnection. Similar problematic
is addressed e.g. by [IOWN].
To provide lossless characteristics, as required by RDMA, there are a
number of technics that are commonly used in DC space, as for example
Data Center Quantized Congestion Notification (DCQCN).
In order to facilitate provisioning of a RDMA compliant slice in the
WAN, a set of attributes shall be described. To be completed in
further versions of this document.
4.6.1. Connectivity attributes
To be completed.
4.6.2. Provisioning procedures
To be completed.
5. Summary of attributes and procedures
After analysing the different use cases, a number of attributes and
procedures can be identified to provide IETF Network Slice services.
Following sections summarize the findings per SLO, SLE and
procedures.
Editor Note: this summary is yet under review.
5.1. Summary of SLOs
The following SLOs can be considered common to the majority of use
cases.
* Bandwidth (or throughput), as an indication of the amount of
traffic allowed to the delivered. It can be expressed
unidirectional or bidirectional.
* Latency, as an indication of the maximum delay expected in a
connection.
Contreras, et al. Expires 27 April 2023 [Page 25]
Internet-Draft IETF Slice Service Interface use cases October 2022
* Jitter (or delay variation), as an indication of the maximum
variation on the delay expected in a connection.
* Packet loss, as an indication of the bounded limit of packet
losses allowed in a connection
* To be completed
5.2. Summary of SLEs
To be completed.
5.3. Summary of procedures
The following procedures allow to cover the analysed use cases.
* IETF Network Slice provision, including allocation and de-
allocation of the slice.
* IETF Network Slice modification (or update) of an existing
allocated slice.
* Retrieval (or query) of IETF Network Slice status and capabilities
of an existing allocated slice.
* IETF Network Slice reservation, allowing a late instantiation of
the slice.
* IETF Network Slice fault management, permitting the collection of
alarms associated to the IETF NEtwork Slice.
* IETF Network Slice performance management, permitting the
retrieval of performance measurements associated to the IETF
NEtwork Slice.
6. Gaps on related IETF Network Slice efforts
Based on the performed analysis, this section captures gaps
identified so far on related IETF Network Slice documents, like the
framework [I-D.ietf-teas-ietf-network-slices] and the NBI YANG model
[I-D.ietf-teas-ietf-network-slice-nbi-yang].
To be completed in forthcoming versions of the document.
7. Security Considerations
This document does not include any security considerations.
Contreras, et al. Expires 27 April 2023 [Page 26]
Internet-Draft IETF Slice Service Interface use cases October 2022
8. IANA Considerations
This document does not include any IANA considerations
9. Acknowledgments
The authors would like to thank (in alphabetical order) Mohamed
Boucadair, Daniele Cecarelli, Dhruv Dhody, Jie Dong, Adrian Farrel,
Joel Halpern, Greg Mirsky, and Gyan Mishra for their review, helpful
comments and suggestions.
10. References
10.1. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119,
DOI 10.17487/RFC2119, March 1997,
<https://www.rfc-editor.org/info/rfc2119>.
[RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
May 2017, <https://www.rfc-editor.org/info/rfc8174>.
10.2. Informative References
[GSMA] "Generic Network Slice Template, version 3.0", NG.116 ,
May 2020.
[I-D.homma-slice-provision-models]
Homma, S., Nishihara, H., Miyasaka, T., Galis, A., OV, V.
R., Diego Lopez, R., Luis Contreras, M., Jose Ordonez-
Lucena, A., Martinez-Julia, P., Qiang, L., Rokui, R.,
Ciavaglia, L., and X. D. Foy, "Network Slice Provision
Models", Work in Progress, Internet-Draft, draft-homma-
slice-provision-models-02, 4 November 2019,
<https://www.ietf.org/archive/id/draft-homma-slice-
provision-models-02.txt>.
[I-D.ietf-teas-ietf-network-slice-nbi-yang]
Wu, B., Dhody, D., Rokui, R., Saad, T., and L. Han, "IETF
Network Slice Service YANG Model", Work in Progress,
Internet-Draft, draft-ietf-teas-ietf-network-slice-nbi-
yang-02, 11 July 2022, <https://www.ietf.org/archive/id/
draft-ietf-teas-ietf-network-slice-nbi-yang-02.txt>.
Contreras, et al. Expires 27 April 2023 [Page 27]
Internet-Draft IETF Slice Service Interface use cases October 2022
[I-D.ietf-teas-ietf-network-slices]
Farrel, A., Drake, J., Rokui, R., Homma, S., Makhijani,
K., Contreras, L. M., and J. Tantsura, "Framework for IETF
Network Slices", Work in Progress, Internet-Draft, draft-
ietf-teas-ietf-network-slices-14, 3 August 2022,
<https://www.ietf.org/archive/id/draft-ietf-teas-ietf-
network-slices-14.txt>.
[IFA032] "IFA032 Interface and Information Model Specification for
Multi-Site Connectivity Services V3.2.1.", ETSI GS NFV-IFA
032 V3.2.1 , April 2019.
[IOWN] "RDMA over Open APN PoC Reference", IOWN , July 2022.
[MEF] "Slicing for Shared 5G Fronthaul and Backhaul", MEF White
paper , April 2020.
[MEF-70] "SD-WAN Service Attributes and Services", MEF-70 , July
2019.
[O-RAN] "O-RAN Xhaul Transport Requirements 1.0", O-RAN.WG9.XTRP-
REQ-v01.00 , November 2020.
[TS23.251] "TS 23.251 Network Sharing; Architecture and functional
description (Release 16) V16.0.0.", 3GPP TS 23.251
V16.0.0 , July 2020.
[TS28.530] "TS 28.530 Management and orchestration; Concepts, use
cases and requirements (Release 16) V16.0.0.", 3GPP TS
28.530 V16.0.0 , September 2019.
[TS28.541] "TS 28.541 Management and orchestration; 5G Network
Resource Model (NRM); Stage 2 and stage 3 (Release 16)
V16.2.0.", 3GPP TS 28.541 V16.2.0 , September 2019.
Authors' Addresses
Luis M. Contreras
Telefonica
Ronda de la Comunicacion, s/n
Sur-3 building, 1st floor
28050 Madrid
Spain
Email: luismiguel.contrerasmurillo@telefonica.com
URI: http://lmcontreras.com/
Contreras, et al. Expires 27 April 2023 [Page 28]
Internet-Draft IETF Slice Service Interface use cases October 2022
Shunsuke Homma
NTT
Japan
Email: shunsuke.homma.ietf@gmail.com
Jose A. Ordonez-Lucena
Telefonica
Ronda de la Comunicacion, s/n
Sur-3 building, 1st floor
28050 Madrid
Spain
Email: joseantonio.ordonezlucena@telefonica.com
Jeff Tantsura
Microsoft
Email: jefftant.ietf@gmail.com
Hidetaka Nishihara
NTT-East
Email: hidetaka.nishihara1104@gmail.com
Contreras, et al. Expires 27 April 2023 [Page 29]