Transcript flexibility of networks

Towards more flexible networks
-- backyard of IMT-2020 -Takashi Egawa
NEC Corporation
Rapporteur, Q.14, SG13
1
From hardware to software
HW world
Dedicated appliances +
Dedicated wire/radio
SW world
Virtual functions +
virtual links
on generic server /
storage / network pool
This shift is expected to contribute to
 Flexible network function
implementation/ operation
 Opex / capex reduction
 Faster business cycle, rapid adaptation
to demand, increased resource usage, …
2
A key difference btw. 5G & 4G:
flexibility of networks
Flexibility is critical in future networks because
 network is becoming even more complicated
multiple RATs, wandering servers/applications, …
 we need slice
Human loves territory -- it simplifies life
 network has to adapt to various changes,
e.g., demand
Internet made changes faster, and made people’s
temper shorter
 we can
Thanks, Moore’s law
3
NW becomes more complicated
 Multi-RAT for
Server
IP networks
QoE Policy
Information
RAN 1
RAN 2
QoE-aware
Scheduler
RAN 3
QoE-aware
Scheduler
QoE-aware
Scheduler
QoE Context
Information
LTE
WiFi
UMTS
LTE
WiFi
WiFi
Inter-RAT
QoE-aware
Universal
Access
WiFi
Inter-RAT
QoE-aware
Universal
Access
better QoE
 Applications /
virtual servers
wanders
 Various reqs
on mobility,
security, …
 Underlying NW
must be flexible
An example of more complicated network architecture:
QoE enhancement in a multi-RAT environment
ITU-R M.2320-0 (11/2014)
4
Network softwarization view
of the 5G mobile networks
ITU-T FG IMT-2020 report (2015/12)
5
Issues to standardize





















Efficient accommodation of various applications
Support for emerging network architectures
Horizontal extension: End-to-end slicing
Vertical extension: Deep data plane programmability (data plane enhancement)
Considerations for applicability of softwarization
End-to-end reference model for scalable operation
Coordinated APIs
Energy management aspects of network softwarization
Economic incentives aspects of network softwarization
Network management and orchestration
Support enhanced MEC management
Support inter-edge mobility of a MEC system
Support more simple and controllable APIs of a MEC system
Support traffic routing among multiple MEC applications
Distributed cloud for service provider
In-network data processing
Resource usage optimization
Resource abstraction
Migration towards newly emerging network
RAN virtualization and slicing under software control
Capability Exposure
ITU-T FG IMT-2020 report (2015/12)
6
A few key questions
 Applicability
Is it really possible to satisfy requirements?
 Copper is inflexible, but reliable in transferring
packets. No worry for ‘QoS’ or ‘priority’
 Dedicated server + real-time OS is good for realtime operations and to guarantee ‘xx msec delay’ ?
 Don’t we really need ‘dedicated hardware’?
 Management
 All the burden of complexity falls in software.
 Necessity for orchestrated operation is increasing.
 Extensibility and exposed capabilities
 how much flexibility should each slice provide? E.g.,
non-IP protocol should be able to be supported?
 How much e.g., API, security capability should be
exhibited to customers?
7
Standardization activities
in/outside of ITU-T
 ITU-T SG13
 SDN, network virtualization, ICN under ‘future
networks’ umbrella
 FG IMT-2020 for 5G
 ITU-T SG15
 Transport SDN
 ETSI Network Function Virtualization (NFV)
 ETSI Mobile Edge Computing (MEC)
 Other activities in other SDOs
TMF Zoom: management and orchestration, IETF/IRTF:
various protocols, 3GPP: RAN virtualization and others
 Open source activities
 OpenStack, OpenDayLight, OPNFV, Opensource
MANO, and many many others esp. in cloud and SDN
8
Y.3011: Network Virtualization for FNs
 Definition of “network virtualization":
 A technology that enables the creation of logically isolated network partitions over
shared physical networks so that heterogeneous collection of multiple virtual
networks can simultaneously coexist over the shared networks. This includes the
aggregation of multiple resources in a provider and appearing as a single resource.
 Motivation
Various Services
LINP3
Virtual
Networks LINP1
LINP2
goals
Physical NW 1
Physical NW 4
Detailed requirements are in Y.3012, architecture in Y.3015
Physical NW 2 Manager
Physical NW 2
Physical NW 3
Physical NW 1 Manager
Isolation, network abstraction,
topology awareness and quick
reconfigurability, performance,
programmability, management,
mobility, wireless
Physical NW 3 Manager

Physical NW 4 Manager
Coexistence of multiple networks
Simplified access to resources
Virtual
Flexibility in provisioning
Resources
Evolvability
Design goals
Virtual Resources
Manager





LINP3 Manager
 Problem spaces & design
LINP1 Manager
 Diverse services
 Heterogeneous network
architectures
LINP2 Manager
 Key technology for Service
Awareness of FNs
Y.3033: Framework of data aware networking
 Definition of “data aware networking":
 A network architecture whose technology optimizes handling enormous
amount of data efficiently in a distributed environment and enables users
to access desired data safely, easily, quickly, and accurately, regardless of
their locations.
 Motivation and scope
 Describing high-level requirements (framework document) to
realize the “data access” design goal specified in Y.3001 which is.
 Optimal and efficient handling of huge amount of data, and
retrieval of the data promptly regardless of their location.
 Problem spaces
 Scalable and cost-efficient
content distribution.
 Mobility support.
 Disruption tolerance.
 Design goals
 Naming, Routing, Caching,
Security, Mobility, API,
Transport.
① On-path caching while downloading content files.
② Responding to user requests from any DAN element.
③ Optimizing process on DAN element before responding.
10
Y.3300: Framework of software-defined
networking
 Definition of “software-defined networking”
A set of techniques that enables to directly program,
orchestrate, control and manage network resources, which
facilitates the design, delivery and operation of network
services in a dynamic and scalable manner
 Objective of SDN
Faster network business cycle, Acceleration of innovation,
Rapid adaptation to demand, Increase in resource availability
and usage efficiency, Customization of network resources
including service-aware networking
 Capability of SDN:
Programmability, Resource abstraction
 Requirements of SDN (a few):
programmability of network
resources, orchestration of network
resources and SDN applications,
logically centralized control of
network resources, abstraction of
underlying network resources
Conclusion
 Softwarization of networks makes
network flexible
--- a key capability of 5G
 Many things to do to realize this
capability, including standardization
 ITU will contribute to this trend
12