1. connect communicate collaborate Green Networking GEANT Robert Pekal Andrew Mackarel TNC 2012

2. connect communicate collaborate Background The energy consumption of the Internet is exploding increasing number of devices connected to it continuous expansion of networks explosion of traffic volume fuelled by video and mobile data applications ICT sector accounts for more than 2% of global carbon emissions. Energy expenses constitute nearly 25% of the total network costs. Carbon footprint restrictions with increased emphasis on corporate responsibility. The energy consumption of the Internet will be one of the main challenges that technology will have to face in the future. First step is to measure – Then, act to reduce consumption and emissions

3. connect communicate collaborate GÉANT Green Team – GHG Audit Strategy Measuring and monitoring GHG emissions are essential features of a strategy to reduce such emissions. The GÉANT Green Team has been carrying out audits of the greenhouse gas (GHG) emissions of NRENs and the GÉANT pan-European network. Adopted the ISO 14064 standard for its carbon audits. Carry out direct measurements where possible Independent validation of audit by 3 rd Party FORCENET Uses categories & template developed by UNIC to enable the audits to be carried out in a methodical and consistent manner.

4. connect communicate collaborate Completed NREN GHG audits NetworkType of Research and Education Network Responsible Partner Year/s GÉANTPan-European (30+ countries)DANTE, UNI-C2009 HEAnetNationalHEAnet2009/2011 HUNGARnetNationalNIIF2009/2011 NORDUnetRegional (5 countries)UNI-C2009/2011 PIONIERNationalPSNC2009 SURFnetNationalSURFnet2010 GRNETNationalGRNET2010

5. connect communicate collaborate GHG audit, categories Office Data centres - services Transportation - commute - on mission Backbone network + Schools network Core networks & data centres / PoPs are the largest factors in network carbon footprints. => where the biggest reductions need to be made.

6. connect communicate collaborate Zero carbon: Tele-working – UNI-C study “SMART 2020” claims:  Enabled by wired or wireless networks, teleworking may save 0,26 Gt CO2e/year out of 51,9 Gt in 2020

7. connect communicate collaborate Zero carbon: Smart 2020 anticipations and claims

8. connect communicate collaborate Zero carbon: Smart 2020 anticipations and claims

9. connect communicate collaborate Tele-working – UNI-C study results UNI-C trail to support „SMART 2020” claim: Public Danish emission factors are used:  Car – 175 g CO2e / km  Bus – 90 g CO2e / passenger km  Regional train – 42 g CO2e / passenger km  Metro – 16 g CO2e / passenger km  319 kg in 5 weeks scales to about 3 tonnes per year for 30 employees  Each Dane emits about 10 tonnes of CO2e per year  3 tonnes is a 1% reduction of the 300 tonnes for 30 employees  The SMART 2020 claim is ½ % of the “business as usual” estimate of year 2020.  UNI-C study thus supports this claim  An extra gain of this study has been to raise the GHG emission awareness of UNI-C’s employees

10. connect communicate collaborate 10 Videoconferencing Pros & Cons Pros You save (travelling) money You save (working) hours You save GHG emissions You save efforts and inconveniences (of travelling) Cons You lose f2f contact with your partners (as well as you lose part of the information!) You lose experiences (of travelling and visiting) and discovering You lose social events and contacts

11. connect communicate collaborate 11 Videoconferencing NIIF Hungary Local Service Summary IP based VC service since 2002 Country wide coverage Infrastructure is (mainly) owned and operated by the NREN Rich service portfolio (MCU, booking, gatekeeper, directory, recording and archiving, streaming, VoIP and PSTN gateways, statistics) Today more than 150 (mostly HD) VC rooms + hundreds of desktop clients (Vidyo) deployed

12. connect communicate collaborate 12 A new service for subscribed users Calculate the working hour requirement and GHG emission caused by a hypothetical f2f meeting (which has been replaced by a real VC) as precisely as possible. A tool has been developed to calculate the above savings and present the results for the users Users can follow their GHG emission and time savings of using of VC on the NREN CRM portal Precise calculation is nearly impossible: it is rather a rough – but conservative! - estimation based on the available data of VC sessions

13. connect communicate collaborate 13 The most important sources of information The central gatekeeper logfiles of the VC sessions (CDR) Location information of VC endpoints (codecs) Route information between parties Speed (time required) and GHG emission of different transportation means

14. connect communicate collaborate 14 VC GHG savings Algorithm Processing the call records of the gatekeeper logfile and save the relevant information in an internal database (done in a separate module, could be tailored to various log formats!) Resolving the location of participants (based on GDS number, IP address or H.323 ID) Calculating the distances Computing the travelling time and GHG emissions Considering the electricity consumption of VC equipments (deduction!) Generating and displaying the results

15. connect communicate collaborate 15 Management Systems - PSNC According to ISO having an environmental policy and establishing a baseline GHG account “may not be sufficient to provide an organisation with the assurance that its performance not only meets, but will continue to meet, its legal and policy requirements. To be effective, they need to be conducted within a structured management system that is integrated within the organisation”. IT systems in large organizations such as Universities are of complex nature and typically utilize many elements to provide the service, such as: Network infrastructure: Network infrastructure: PoE switches, routers, firewalls and others. Servers: Servers: hardware, virtual machines, terminal servers, clusters and others. Storage and backup: Storage and backup: RAID arrays, SAN devices, tape storage and others. Software: Software: databases, application /web servers, user applications (as ERP, CRM, Business Intelligence systems) Voice infrastructure: Voice infrastructure: softswitches, phone system, PSTN gateways and others. Power consumption is an important part of OPEX.

16. connect communicate collaborate 16 Green Steps Power management for unused computers Use more energy efficient computers first Optimize load distribution between servers Replace suboptimal computers Optimize / replace suboptimal software Eliminate underutilized network devices React to abnormal energy consumption

17. connect communicate collaborate 17 Green Process Action Monitoring Decision support Automation

18. connect communicate collaborate 18 Power Management as a decision problem decision making process The problem can be seen as a decision making process based on the current state of environment (in this case consisting of hardware and software elements used to provide the service). sensors Multiple sensors are required to monitor elements of the service (e.g. server load, traffic, temperature, etc.). decision-making agent rules A decision-making agent is required to decide what actions need to be taken based on sensor input, historical data and rules. effectors A number of effectors are required to implement configuration change requests (e.g. move virtual machine image, shut down cluster host, etc.). Service (consisting of hardware and software elements) Agent Historical data Rules Sensors Effectors

19. connect communicate collaborate 19 Predictive Analytical Module Verax Green NMS is equipped with a predictive analytical module allowing for: Non-intrusive (i.e. meter-less) measurements of momentary power consumption of each managed element. Analytical module for a visual presentation of energy consumption per logical blocks of a data-center. The analytical module helps operators to determine the most power-critical elements of their IT service (e.g. databases, network infrastructure, web static content servers, etc.) Detection of abnormal power consumption and generation of power- related alarms based on historical data. Optimization module suggesting reconfiguration actions required to minimize power consumption for existing applications and hardware.

20. connect communicate collaborate 20 Innovation within Green NMS hardware & software Holistic approach (hardware & software levels) Non-intrusive Non-intrusive approach (extra meters not required) - heuristic algorithms where direct measurement impossible complex simulations Advanced multi-criteria decision support system (including complex simulations) Remote Remote energy saving actions automatically Implementing energy saving policies automatically (human involvement reduced)

21. connect communicate collaborate Next steps To recruit more NRENs in environmental policy and practice, promoting the added value of NRENs, and working with others to meet the targets of carbon neutrality. To collect and disseminate best practices and results to research and academic community. To set targets for efficiency and for use of carbon offsets based on the data available on the reports To increase the uptake of services which are based on or which generate carbon reduction effects, such as virtualisation, tele-working, videoconferencing. To further develop NREN insights and practices on Green IT, so that others may benefit. To “push” NRENs to improve energy efficiency in networking and computational infrastructures.

22. connect communicate collaborate Thank You