Bolla-Input Project v1.2 - ETSI EE Worshop 2015 · Devices, usually placed in users’ homes (e.g.,...
Transcript of Bolla-Input Project v1.2 - ETSI EE Worshop 2015 · Devices, usually placed in users’ homes (e.g.,...
INPUT
Pursuing energy-efficiency in ICT through virtualization:
the INPUT Project Prof. Raffaele Bolla
CNIT – University of Genoa [email protected]
Third ETSI Workshop on ICT Energy Efficiency and Environmental Sustainability 3-‐5 JUNE 2015
SOPHIA ANTIPOLIS, FRANCE
Outline § Input Project
ü Basic data ü Objectives ü Architecture
§ Input potential impact on energy consumption the Set-Top Box example
§ The relevance of the interfaces
§ A proposal: Green Abstraction Layer v2.0
INPUT at a Glance
§ Title: In-Network Programmability for next-generation personal cloUd service supporT
§ Project duration ü January 2015 – December 2017 (36 months)
§ Consortium ü 9 partners from 5 countries
§ Project budget ü 3.1 M€ (100% EU funded)
§ Resources ü 511 PM (~15 full time persons over three years)
§ Web-site and contacts ü http://www.input-project.eu ü [email protected] (coordinator), [email protected],
The INPUT Consortium
N. Partner Name NaOon
1 CNIT ü Genoa Research Unit ü Catania Research Unit (including Reggio Calabria
and Cagliari)
2 Ericsson Telecomunicazioni S.p.A.
3 UBITECH
4 Dublin City University
5 HOP Ubiquitous S.L.
6 Infocom S.r.l.
7 COSMOTE
8 Telecom Italia S.p.A.
9 Julius-‐Maximilians-‐Universitaet Wuerzburg
Objectives
§ Designing a novel infrastructure and paradigm to support Future Internet personal cloud services in a scalable and sustainable way and with innovative added-value capabilities.
§ Enabling next-generation cloud applications to go beyond classical service models, and even by replacing physical Smart Devices, usually placed in users’ homes (e.g., set-top-boxes, etc.) or deployed around for monitoring purposes (e.g., sensors), with their virtual images, providing them to users “as a Service.”
§ Virtual and physical Smart Devices will be made available to users at any time and at any place by means of virtual cloud-powered Personal Networks, which will constitute an underlying service model.
§ These Personal Networks will provide users the perception to be always in their home LAN with their own smart digital (virtual and physical) devices.
§ Fostering future-proof Internet infrastructures that will be “smarter,” fully virtualized, power vs. performance optimized, and vertically integrated with cloud computing, with a clear impact on OPEX and CAPEX of Telecoms, of Service Providers, and of end-users.
Main Approach
§ The INPUT project will overcome current limitations in the cloud service design due to the underlying obsolete network paradigms and technologies, by: ü using the computing and storage capabilities of network
appliances to allow users create private clouds “in the network”;
ü exploiting the energy consumed in network appliances more efficiently than in traditional cloud computing scenarios;
ü moving cloud services much closer to end-users and smart-devices, thus avoiding useless network infrastructure overloading.
§ Extending the programmability of network devices to make them able to host cloud service applications, which will cooperate with the ones in users’ terminals and datacenters to realize the aforementioned cloud services.
Main Approach
Personal Network
Virtual Smart Devices
ü Personal Networks provide users with the percep[on of always being in their home Local Area Network with their own (virtual and physical) SDs, independently of their loca[on.
ü Personal Network service roughly corresponds to the Telco-‐based virtualiza[on of HomeGateways.
Reference Architecture
End-‐user
Data Center
Service_Apps
DC_Apps
Edge Network Device
Personal Network
Virtual Smart Devices
Home Network
User_Apps NBI
Network and Service Mgmt SBI (SDN/NFV) In-‐Network
Programmability
An estimation of the INPUT impact on energy consumption: the example of set-top boxes
§ We have considered the potential effects of three different actions that are exploited in the Input scenario for what concern the Set-Top Box (STB) consumption: ü Virtualization: moving objects from user’s home
to servers in Telco networks/datacenters ü (Power Management) Dynamic Adaptation:
dynamic variation of energy consumption for servers, from a minimum to a maximum value
ü (Power Management) Consolidation: Some servers can be set in stand-by and the total load can be concentrated in a minor number of servers
BAU scenario Device τ [y] ECF
[kgCO2 eq] Wh/y [kWh] (3) OCF1y [kgCO2 eq] Ownership [n]
TCF1y [MtCO2 eq]
Set-‐Top-‐Box 3.90 (1) 25.00 (2) 107.89 (1) 52.85 0.52 (3) 3.27
(1) h^p://www.energyra[ng.gov.au/wp-‐content/uploads/Energy_Ra[ng_Documents/Library/Standby_Power/Standby_Power/E3-‐2010-‐Intrusive-‐Survey-‐FINAL-‐Report.pdf (2) h^p://onlinelibrary.wiley.com/doi/10.1111/jiec.12145/epdf (3) h^p://remodece.isr.uc.pt/downloads/REMODECE_PublishableReport_Nov2008_FINAL.pdf
INPUT Scenario Server τ [y] ECFS [kgCO2 eq] Power [W] Wh/y [kWh] OCF1y [kgCO2 eq] TCF1y [kgCO2 eq]
Mid-‐range 5 360 (1) 607 (2) 13525.67 (2) 8115.40 8187.40
(1) h^p://pubs.acs.org/doi/ipdf/10.1021/es303012r (2) h^p://www.sciencedirect.com/science/ar[cle/pii/S0140366414000619
TCF1YS = NS ∗
ECFS
τ S +OCFV =NHH ∗OWN
nSTB
"
##
$
%%∗ECFS
τ S +OCFV
!"#!!! = 0.6 ∗ 24 ∗ 365 ∗ !!"#! !
!! =!!!×!"#
!!"# !
• ConsolidaOon effect (from NS to M ac[ve servers)
• Dynamic AdaptaOon effect
• VirtualizaOon
Comparison
-‐100
-‐80
-‐60
-‐40
-‐20
0
20
40
60
80
100
25 75 125 175 225 275 325 375 425 475 525 575 625 675 725 775
Carbon
Foo
tprin
t Saving [%
]
nSTB [#]
V V + DA V + DA + C
INPUT – Virtualiza[on + Dynamic Adapta[on + Consolida[on
ρ Ns M TCF1y [MtCO2 eq] TCF Saving [%]
4/24 343200 113256 (33% N) 0.59 82.4
BAU 38 kg CO2 eq
INPUT 7 kg CO2 eq
V: virtualiza[on effect DA: Dynamic Adapta[on effect C: Consolida[on effect nSTB= number of STBs on a single server
Working Layers & Interfaces
Service_Apps, SLAs, user accoun[ng, data, etc. Northbound Interfaces
Personal Network Provider
Southbound Interfaces
Cloud Service Providers
ConsolidaOon
OrchestraOon Monitoring
Flexible Network PrimiOves «in Network» Programmable Resources
Network and Service
Management
Config of Network Services Management of Service_App
Measures o
n Users’ Q
oE
and on
network/Ap
ps QoS
Personal Network
Cloud Service Providers Cloud Service Providers
Personal Cloud and Network Services
Infrastructure/Plaform Provider
Add/remove service, configure components, etc.
GAL
Green Abstraction Layer § The GAL is a hierarchical
interface to control and to orchestrate power management primitives in a network device in a scalable and flexible way.
§ The GAL layers allow to divide and conquer the complex process of optimizing the mapping between power management primitives (acting at the HW level) and the network logical/virtual/ functional configuration.
§ The GAL has been approved on March 2014 as ETSI Standard 203 237.
The definition of interfaces is crucial in this scenario. We strongly advocate to extend the current standard interface (GAL) to support the dynamical “energy consumption” management of virtual functions/objects (i.e., NFV), as well: GREEN ABSTRACTION LAYER v2.0
Thank you for your kind attention Any Questions?