EIT ICT Labs Smart Network at the Edge

EIT ICT Labs
Smart Network at the Edge
Antonio MANZALINI (Telecom Italia)
EIT Activity AITA 13 231
Smart Networks at the Edge
Vision and Challenges
Overall Goal and Objectives
Some Results
Lessons Learnt and Future Outlook
…well known principles…
 SDN: decoupling SW control
plane
from
HW
(data
forwarding), and moving logic
and states to programmable
controllers.
 NFV: virtualizing some network
functions that can run on
standard HW, and that can be
moved
and instantiated in
various locations of the network.
Drivers
HARDWARE
Increase of performance
of standard HW
Costs reductions of IT
processing and storage
systems
Disaggregation
of
IT
resources in components
(e.g., CPU, SSD, HDD)
TERMINALS
SOFTWARE
Growing availability
Open Source SW
of
Transforming
network
functions in SW modules
Emergence
of
ecosystems
around
Open Source SW
More and more powerful
Users’
devices
(e.g.
tablets, any handsets, etc)
Smart
“Things”
embedded
communications
with
Border between terminals
and networks is blurring
Trends
The Network will be more and more applicationdriven, with services and traffic dictated by the endUsers apps dynamics.
Technology adoption
…cell phone took less than 10 years to reach 25%
of the US population while the telephone took over
30 years.
Commoditization
Reducing time to market
Competition
Automation
How finding new revenues ?
Telecom Italia - All rights reserved
Transition to Economy of Knowledge
Traffic growth
How reducing Opex ?
How taming complexity ?
Sustainability
7
Vision: a change of paradigm (1/2)
1. Integrating profoundly Cloud/IT resources and Carriers’
Networks:
 distributed virtual platforms executing any network
function (e.g., L4-L7 or even L2-L7) and services as
“applications” (on VMs, dynamically allocated and
moved on general purpose HW);
2. Blurring the distinction between the “Carriers’ Network”
and what connects to it, i.e., the End-Users “Terminals”:
 any devices, machines, smart things, robots,
drones…will look like nodes (at the edge) providing the
End-Users with “any services”.
Vision: a change of paradigm (2/2)
Core
 potential reductions of CAPEX
and OPEX;
 …need to test the performance
 convergence of IT and Networks
nodes and systems;
 …big impact on operations
processes;
 standardization of interfaces
 … Standards de Facto
 interoperability
with
legacy
equipment;
 development of high-skill jobs for
mastering the software.
Edge
accumulation of processing and
storage resources;
migration of “intelligence” towards
the End-Users;
enabling new ICT ecosystems
 lowering the threshold for new
Players to enter the edge
arena;
 new forms of competition and
collaboration among Players;
 new value chains and new
business models;
Leaving more time for a smooth
“softwarization” of the core.
Overall Goal
 Investigating S/W models, tools and experimental
trusted validation procedures to foster the
development and deployment of Smart Networks
at the Edge.
Examples of results
 Results reported in three main Deliverables
 D1301 - Requirements
 D1302 - Simulation tools
 D1303 - Test beds results
 D1301 Requirements
 Use Cases…
 Personal Data and Services following Users
 Harnessing idle resources at the Edge
 End-to-End Services across Edge Networks
 Security monitoring of mobile devices
 …and related requirements about methods and systems
 e.g., about «management and orchestration» of network functions
executed in VMs
Examples of results
 D1302 Simulations
 Quick tool (in Scilab) for optimising the dynamic allocation of
logical resources across a number of edge networks
 VMs are dynamically allcated and moved to using heuristics
Examples of results
 D1303 Test beds
 proof-of-concept demonstrations (at TI
and UniBO)
 feasibility of an integrated SDN node
capable to provide both connectivity
and network services
 Joint experiment UniBo-TU/e
 A User is provided with a streaming
video through an edge access s/w
router
 …both the video server and the s/w
router are executed on VMs which
migrated at once
 QoE measurement: evaluate the
impact of the VM migration on the
overall QoE of the video (e.g., by frame
comparison)
Lessons learnt and future outlook
 Taming «complexity» = extracting «simplicity»
 defining appropriate “abstractions” (e.g., logical resources)
 simplifying the “network views” (e.g. in terms of slicing)
 limiting the number of network “touch-points”
 Management and Orchestration of VMs
 automating Operational processes (reducing time to
market)
 managing the life cycle of VMs (creation and deletion)…but
also installation, configuration, monitoring and exec. of
apps in the VMs
 Elasticity and flexibility
 VMs placement, move and traffic routing between VMs
(e.g., double constrained optimization problems to be
solved real time)
Research Agenda
Key areas will include:
oNew Management and Orchestration approaches
integrating abstractions;
o QoS vs QoE;
oDeveloping and controlling “intelligence” into End-Users
devices, machines, smart things…drones, robots;
oStandardization of interfaces for interoperability;
oAddressing Security and Privacy;
oDeveloping New Business Models and Ecosystems;
oHow pursuing Open Source Hardware and Software;
oEducation and Development of new skills.
A vision of the future
Antonio Manzalini – Confidential Telecom Italia - All rights reserved
networking, transmission
Core
Data Centers (Cloud)
• Computing for Global
Intelligence
Acces
s
Edge Mini Data Centers
• Computing for Edge
Intelligence
X As A Service
• Sensors
• Actuators
• Computing for
Local Intelligence
Requirement: Ultra Low Latency
to get milliseconds of reaction time
Lessons learnt and future outlook
IEEE SDN Initiative
a number of committees are exploring and
developing conferences, education modules,
standards, publications, proofs of concepts, preindustrial adoption of SDN.
Arrivederci !
[email protected]