ITU-T Focus Group on IMT2020 Workshop on Network Softwarization 21st · 2015-09-21 · ITU-T...

ITU-T Focus Group on IMT2020 Workshop on Network Softwarization

21st September 2015, Turin

Prof. Alex Galis http://www.ee.ucl.ac.uk/~agalis/

Communications and Information Systems Group Department of Electronic & Electrical Engineering

University College London, United Kingdom

Challenges in 5G Network and Service

Softwarization & Infrastructures

ITU-T IMT2020 5G Network Softwarization Workshop – Turin 21st September 2015

1. Internet - remarks & lessons for Network Softwarization 2. Towards Network Softwarization – remarks & lessons from the past 3. 5G Software Networking Architecture & Towards a new Network Softwarization Model 4. Some Research & Development Challenges 5. Early Network Softwarization Developments @ UCL 6. 5G Research Projects 7. Conclusions

Contents


Rough sketch by Larry Roberts ARPAnet Plan – late 1960s


Internet 1973-74

UCL connected in July 1973 to ARPAnet


Current Internet (including 3 & 4G networks)

•  The Internet as a connectivity platform plays a central and vital role in our society

! Work and business, education, entertainment, social life, …

•  Victim of its own success, suffering from ossification

! Innovation meets natural resistance (e.g. no IPv6, no mobile IP, no inter-domain DiffServ, no inter-domain multicast, etc.)

•  Services such as P2P, IPTV, Cloud services, emerging services, pose new requirements on the underlying network architecture. OPEX costs are up to 90%

•  Big growth in terms of the number of inter-connected devices but slow growth in new services


Key Changes in Internet - History

•  Changes were possible when the Internet was still an academic research network (i.e. until 1993 when the WWW turned it to a commercial)

•  Inter-network that underpins the “information society”

•  Key changes in that period were the following: 1982 DNS, 1983 TCP/IP instead of NCP, 1987 TCP congestion control, 1991 BGP policy routing, 1991 SNMP

•  No significant changes since then apart from MPLS which has been deployed in addition to plain IP

•  Research efforts towards the Future Internet: evolutionary & clean-slate approaches, autonomic management, Internet softwarization


4G Network Architecture


Some current systemic limits in Networking

"  Networks are becoming both a connectivity and service execution environment

"  Computation, storage and connectivity Virtualised separately (but not in an integrated way)

"  Silos and disparate systems with limited extensibilities which created a segmentation of networking & computation

"  Need for Software driven / enabled features:

àWork towards a service and management aware connectivity infrastructure #  Work towards a flexible and cost effective integrated virtual

infrastructure with elastic usage and sharing resources

#  Programmability: dynamic and autonomic activation of network and service functions

•  Programmability and Elasticity •  Integrated Virtualisation of Connectivity Storage and Processing Resources •  In-Network Management •  Service awareness

•  Energy awareness •  Content awareness •  Knowledge awareness •  Economic awareness •  Extensibility with new features •  ............

à


Drivers for Change to 5G Networking •  Disappearance of the ‘End-host only’ concept ( i.e. edge networks;

mobile edge computing, new nodes : sensors, mobile devices; ) •  Lack of in-system management (i.e. information, decision,

implementation – closed control loops for realizing management requirements)

•  Trustworthy User / Network / Service (i.e. end-host protocols can and are altered # many security issues)

•  No explicit media & content handling •  Size & Costs:

•  N X 109 connectivity points - status: reaching maturity and maybe some limits •  N X105 services /applications - status: fast growing •  N X103 Exabyte's content - status: fast growing •  Cost structure: 80% (#90%) of lifecycle costs are operational and

management costs - status: reaching crisis level •  Ossification: reaching crisis level

•  A lot of missing and interrelated features; missing enablers for integration and orchestration of Nets, Services, Content, Storage

•  Substantial barriers to innovation with novel services, networking systems, architecture and technologies


How to Change

Approaches: •  Parallel Networking; Progressive changes; “Cleaner” slate and

evolutionary •  Network Virtualization •  Virtualization of most Networking Functions •  Network of networks # System of coordinated service networks

(Network & Service Orchestration) •  Network control plane # Service & Network control plane •  Virtualization of resources (Connectivity /Networks, Computation,

Services, Content, Storage, Devices) •  Programmability at all levels •  Increased self/autonomic in-management as the means of

controlling the complexity and the lifecycle costs •  Softwarization of Networks



Contents


P1520 Reference Model –Application Programming Interfaces [i.e. Dynamic Service Chaining (Service Deployment Concept)~ 2004]

Applications invoking methods on objects below

UInterface

CustomisedRouting

RoutingAlgorithms

RSVP or Other per-flow protocol

Policy-BasedDifferentiated

ServicesScheduling

LInterface

Service-specificbuilding blocks

Resourcebuilding blocks

Basebuilding blocks

Deg

ree

of A

bstr

actio

n

High

Low

VInterface

USERS

CCMInterface Controller

Hardware and other resources Routing table Data

Remarks: 1. P1520 has no hosting environment(s) for the network services # dynamic service chaining and the evolution of network virtualization from data centers into carrier networks do not come without their own challenges.


Programmable Network Model ( ~ 2005)

Network API

Node API

Applications

Execution Environments

Node

EE1 EE2 EEn

Router

Node OS

Remarks: 1. Focus on “low level” programming the network; 2. Virtualisation of networks via programming of networks 3. Extremely hyperactive network which would be difficult to manage # Needing programming network services (instead of re-architecting the network and OSs for every service)


Network API

Node API

Applications


Node

EE1 EE2 EEn

Router

Node OS

Softwarization Architecture Connectivity & Computation Infrastructure

Status in the early 2000+ ( active & programmable networks)

Softwarization Evolution


Draft 5G Software Networking Architecture Applications & Services Plane

Service & Network Control & Execution Plane

Forwarding Plane

Software Platform - Virtualised infrastructure

User Equipment & Devices

Mobile Fronthaul Routers Back Fronthaul IoT sensors Network/ Connectivity

resources

Radio Access Technology

Computation & Storage Resources Servers Micro & large scale Data Centres Storage & Computation resources Evolved

Packet Core

Physical infrastructure (connectivity, computing, storage and IoT resources)

Mobile Core NFV

Radio access network (RAN) NFV

UE NFV

Service Programmability & Multi-domain Orchestrations

Slice A

Slice B

System m

anagement

Clouds

App-Driven API API

Data &

Control

Data &

Control

Data &

Control

Data &

Control

Data &

Control

Mobile Edge Computing

Data &

Control

Domain Orchestrator

Domain Orchestrator

Domain Orchestrator

Domain Orchestrator

Domain Orchestrator



Contents


Northbound Service APIs

Governance and Business Applications Plane

Applications Execution Environments

Network Node

Processing Node

Network Device

Network Device

Physical Infrastructure

Northbound Control APIs

2. Controllers for Platform Functions and Services Chains

(i.e. Placement, Orchestration, etc)

Virtual Service Control Plane

1. Platform Execution Environments

Control APIs

3. Controllers for Resource Functions and Services

(i.e. Middlebox, Connectivity, Routing)

Software Virtualised Infrastructure

1. Resources Execution Environments

• Control Elements for Resources & Integrated Virtualisation, • Networked hypervisors • Rapid VMs movements, Virtual networks, Network Function Virtualization

• Dynamic networked environments that combine virtual networks with virtual compute nodes within the same topology • Control Elements for Resources & Integrated Virtualisation, Networked hypervisors • Rapid VMs movements, Virtual networks • Control Elements for VMs •Creation of groups of VMs, Orchestration, multi goal optimisation, Resources Virtualisation Functions, Virtual Machines management (QoS, rapid movement, monitoring, federation) , VMs Information Base, Service-awareness Enablers, Execution Environments Management, Network Services, Self-management Functions, OPEX Management

Governance, Service Orchestration, Deployment, execution, global manifest

Softwarization Revised Model ( ~ now)


Key Challenges – Future 5G Networking Systems

(Self) Autonomic Management

Service Control Mechanisms

Network and Computing Operations

Service wireline environments

Service wireless environments

Orchestration & Coordination

Interoperability and Federation

Mechanisms and methods for communication and programmability of service modules

Authentication for service modules

Multi-domain Orchestration

Unification, Adaptive and Autonomic Methods and Systems

Optimised allocation and orchestration of resources

Energy management and optimisation

Security and Safety and Scalability

Efficient NFV & Network Virtualization & Slices

Programmability methods and mechanisms

Performant and Safe Network Execution environments

Service programmability and service manifests

Lightweight Execution Environments

System Hosting

Architecture & Integration &

Models



Contents


•  It is composed of platforms and systems that are Open Source and are actively under development (http://clayfour.ee.ucl.ac.uk).

•  It allows us to build distributed dynamic networked environments that combine virtual networks with virtual compute nodes within the same topology.

•  The main elements include: •  A virtual environment using hybrid resources ( e.g. network &

computation resources) •  A New Network Hypervisor •  Various Placement engines, for placing virtual elements •  A mechanism to setup experiments for network functions •  Autonomic Management tools for the above systems •  Monitoring Framework for the above systems •  Orchestration functionality •  Information system & platform specific to the above platforms

Service Defined Systems - Open Source TestBed


Pla

cem

ent E

ngin

e

Orc

hest

ratio

n

Hos

t Con

trolle

rs

Mon

itor P

robe

s

Run

time

Eng

ine

VR

Pro

toco

l Sta

ck

VR

App

licat

ion

Env

ironm

ent

Rou

ter C

onfig

urat

or

Link

Con

figur

ator

Topo

logy

Con

figur

ator

Mon

itorin

g E

ngin

e

Con

figur

atio

n A

ctua

tors

Info

rmat

ion

Sto

rage

&

Inde

xing

Info

rmat

ion

Flow

E

stab

lishm

ent

Info

rmat

ion

Pro

cess

ing

&

Agg

rega

tion

Kno

wle

dge

Pro

duct

ion

Info

rmat

ion

Dis

sem

inat

ion

UC

L S

ervi

ce N

etw

ork

Sof

twar

izat

ion

Test

-Bed

- S

/W

Sta

ck F

unct

iona

lity

Physical Hosts

VM

for

Virtu

al R

oute

r Fu

nctio

nalit

y

Virtu

al L

ink

Fu

nctio

nalit

y

Platform to Manage the Virtual Infrastructure - Dynamic networked environments that combine virtual

networks with virtual compute nodes within the same topology (i.e., creates,

monitors, configures, manages virtual networks & runs VIM scripting applications)

Management of Network VMs (i.e., creates, monitors, configures and runs VMs for the network: virtual routers, virtual links &

VR applications)

Platform to Manage Information / Knowledge in the Virtual Infrastructure (i.e.,

collection, dissemination, storing, optimisation, aggregation, information flows

establishment / optimization)

VIM

Con

trolle

r

Scr

iptin

g E

ngin

e

Virtual Infrastructure Management (VIM) In

form

atio

n Fl

ow

Opt

imiz

atio

n

VM

for

App

licat

ion

Func

tiona

lity

Lightweight Network Hypervisor (LNH)

Info

rmat

ion

Col

lect

ion

Information & Knowledge Management (IKM)

Client Management Applications / Network Services

Service Defined Networks – UCL Open Source TestBed


Energy Management www.dolfin-fp7.eu


5G Orchestration www.5g-ppp.eu/sonata

.A BCA C (A F A

AC

&CA

BDC

) C AB C A B

.A

.A

)

A BCAD CDA

C A C A C

A - CF A (D C

C

CA

) (

C

DB

CB

CG

DC

AA

DC.

CI

AA

&

.A

) C AC A

A BCAC A

&C A

B C AC A

-(

.A

DA

��

��

��

.A )

EU 5G-PPP project: SONATA - Service Programing and Orchestration for Virtualized Software Networks (Single Domain)

www.5g-ppp.eu/5gex

EU 5G-PPP project: 5G Exchange (5GEx) - Multi-domain Orchestration for Software Defined Infrastructures



Contents


Draft 5G Software Networking Architecture

Mobile Core NFV

Radio access network (RAN) NFV

UE NFV

Service Programmability & Multi-domain Orchestrations

Slice A

Slice B

System m

anagement

Applications & Services Plane

Clouds

Service & Network Control & Execution Plane App-Driven API API

User Equipment & Devices

Mobile Fronthaul Routers Back Fronthaul IoT sensors Network/ Connectivity

resources

Radio Access Technology

Computation & Storage Resources Servers Micro & large scale Data Centres Storage & Computation resources

Software Platform - Virtualised infrastructure

Data &

Control

Data &

Control

Data &

Control

Data &

Control

Data &

Control

Evolved Packet Core

Mobile Edge Computing

Data &

Control

Physical infrastructure (connectivity, computing, storage and IoT resources)

Forwarding Plane

Domain Orchestrator

Domain Orchestrator

Domain Orchestrator

Domain Orchestrator

Domain Orchestrator


Concluding Remarks 5G Networks are both a connectivity and service execution environments Network Softwarization, Service & Network Control Plane facilities and Self & Autonomic Management and Service would represent nearly 99% of the new 5G Networking functionality !!! Why now: •  Virtualisation and programmability are cost effective

and operational •  Continuous demands for large number of software

features and qualities •  Accelerating growth in Services & Applications


Some Relevant References

•  Galis, A., Denazis, S., Brou, C., Klein, C.–”Programmable Networks for IP Service Deployment” ISBN 1-58053-745-6, pp 450, June 2004, Artech House Books, http://www.artechhouse.com/International/Books/Programmable-Networks-for-IP-Service-Deployment-1017.aspx

•  Nakao, A. -"Deeply Programmable Network", Emerging Technologies for Network Virtualization, and Software Defined Network (SDN), ITU-T Workshop on Software Defined Networking (SDN), http://www.itu.int/en/ITU-T/Workshops-and- Seminars/sdn/201306/Documents/KS-Aki_Nako_rev1.pdf.

•  Y.3001 ITU-T recommendation – “Future networks: Objectives and design goals” – July 2012 @http://www.itu.int/rec/T-REC-Y.3001-201105-I

•  A. Galis et al., “Softwarization of Future Networks and Services – Next Generation SDNs” @ Proc. of IEEE SDN4FNS’13, Trento, Italy, Nov. 2013 @http://sites.ieee.org/sdn4fns


Thank You


Spare Slides


Network API

Node API

Applications


Node

EE1 EE2 EEn

Router

Node OS

SDNs Architecture Connectivity & Computation Infrastructure

Status in the early 2000+ ( active & programmable networks)

Softwarization Evolution - Conceptual Networked Systems


Future Networks : Objectives Vs. Design Goals

Y.3031 - IDconfig

Y.3011 - FNvirtualisation

Y.3021 - FNenergy

Y.3001 - FNobjectives&designgoals


Software Defined Networks - 12 Design Goals

①  (Service Diversity) FNs should accommodate a wide variety of traffic and support diversified services

②  (Functional Flexibility) FNs should have flexibility to support and sustain new services derived from future user demands

③  (Virtualization of resources) FNs should support virtualization so that a single resource can be used concurrently by multiple virtual resources.

④  (Data Access) FNs should support isolation and abstraction FNs should have mechanisms for retrieving data in a timely manner regardless of its location.

⑤  (Energy Consumption) FNs should have device, system, and network level technologies to improve power efficiency and to satisfy customer’s requests with minimum traffic

⑥  (Service Universalization)FNs should facilitate and accelerate provision of convergent facilities in differing areas such as towns or the countryside, developed or developing countries


Software Defined Networks - 12 Design Goals (Cont.)

⑦  (Economic Incentives) FNs should be designed to provide sustainable competition environment to various participants in ecosystem of ICT by providing proper economic incentives

⑧  (Network Management) FNs should be able to operate, maintain and provision efficiently the increasing number of services and entities.

⑨  (Mobility) FNs should be designed and implemented to provide mobility that facilitates high levels of reliability, availability and quality of service in an environment where a huge number of nodes can dynamically move across the heterogeneous networks.

⑩  (Optimization) FNs should provide sufficient performance by optimizing capacity of network equipments based on service requirement and user demand.

11   (Identification) FNs should provide a new identification structure that can effectively support mobility and data access in a scalable manner.

12   (Reliability and Security) FNs should support extremely high-reliability services


1. Service Diversity 2. Functional Flexibility 3. Virtuallization /resources 4. Data Access 5. Energy Consumption 6. Service Universalization 7. Economic Incentives 8. Network Management 9. Mobility 10. Optimization 11. Identification 12. Reliability & Security

34

ENVIRONMENT AWARENESS

SERVICE AWARENESS

DATA AWARENESS

SOCIAL-ECONOMIC AWARENESS

Future Networks : Objectives Vs. Design Goals


Technologies - achieving the design goals

•  Virtualization of Resources (Network Virtualization) •  Enables creation of logically isolated network partitions over

shared physical network infrastructures so that multiple heterogeneous virtual networks can simultaneously coexist over the shared infrastructures; it allows the aggregation of multiple resources and makes the aggregated resources appear as a single resource

•  Data/Content-oriented Networking (Data Access) •  Energy-saving of Networks (Energy Consumption)

•  Forward traffic with less power •  Control device/system operation for traffic dynamics •  Satisfy customer requests with minimum traffic

•  In-system Network Management (Network Management) •  Distributed Mobile Networking (Mobility) •  Network Optimization (Optimization)

•  Device / System / Network level optimization (Path optimization, Network topology optimization, Accommodation point optimization) 35


SDN Architectural Model (Source ONF ~ 2014)

Remarks: 1. industry acceptance of management & control & data planes decoupling 2. underdeveloped service & management planes


NFV Architectural Model (Source ETSI ~ 2014)

Remarks: 1. Virtualisation of some network appliances / middleboxes based (network) functions #Retrofitting programmability of networks / services means substantial architectural changes # Needing programming network services (instead of re-architecting the network functions and OSS for every service)

ITU-T Focus Group on IMT2020 Workshop on Network Softwarization 21st · 2015-09-21 · ITU-T...

Documents

Transcript of ITU-T Focus Group on IMT2020 Workshop on Network Softwarization 21st · 2015-09-21 · ITU-T...