Newsletter‎ > ‎

Newsletter - Volume 3 Issue 3 - December 2014

Monthly bulletin of the IEEE Computer Society Special Technical Community on Sustainable Computing

Providing quick access to timely information on sustainable computing.

Sponsored by:





From the Chair's Desk

by Danilo Ardagna, Politecnico di Milano

I’m happy to open the December issue our e-letter.  The new chairs  that have joined us this year are putting a lot of effort on this and we are able to publish quarterly  very interesting material. I thanks Diwakar for coordinating this and all Information officers for their contributions. 
Moreover, I’m happy to announce that our site includes a Webinar section.  The first video is form Cristina Rottondi
(On the Complexity Electric Vehicles Recharge  Scheduling).  I’ll be happy if all the members would like to participate and  provide input for  this.  Our STC can become a place were we can share new ideas and foster discussion on our STC related topic.   In case you are interested in, please contact me to danilo.ardagna@polimi.it

Message from the editor

Diwakar Krishnamurthy, University of Calgary

I'd like to once again thank our hardworking team of information officers for bringing out this issue.

Dr. Cristina Rottondi continues her series of articles on issues pertaining to the smart grid.
Dr. Burak Kantarci and Dr. Massimo Tornatore discuss greening of data centers and cloud computing systems in separate articles. Finally, in our conference showcase Dr. Patricia Lago summarizes the first two editions of the ICT for sustainability conference that she helped co-organize. As always, I encourage you to get back to us with comments, feedback, and contributions.


Demand Management of Residential and Commercial Users: Techniques and Approaches

by Cristina Rottondi, Politecnico di Milano

https://sites.google.com/a/ieee.net/stc-sustainable-computing/Newsletter/newsletterjul14/DSC01091.jpgResidential and commercial users are expected to play a key role in improving the efficiency of power grids
since they are among the major contributors to the global energy balance. However, the consumers' demand
has been so far largely uncontrollable and inelastic with respect to the power grid conditions. To address this
issue, demand management mechanisms can be used to optimally manage the electric resources of users, with
the aim of not only reducing their bills or saving energy, but also using more efficiently the energy itself by
means of shifting loads to off-peak hours, adapting the demand to renewable sources supply or reacting to
emergency conditions [1].
Demand management frameworks are designed to optimally manage electric resources of users through a
specific architecture, shown in Figure 1, composed of the following basic components:


  •  Local generators: are local energy plants (e.g., photovoltaic (PV) plants), which generate electric energy
    that can be either used locally or injected into the grid.
  •  Smart devices: are electric appliances which are able to monitor themselves and that can be remotely controlled.
  •  Sensors: are used to monitor several data of interest, such as the user's position within the house, temperature and light.
    Moreover, in the case of traditional devices, power meter sensors can be used tomonitor and control these appliances.

  •  Energy storage systems (ESSs): are storage devices which allow the system to be flexible in managingelectric resources.

  •  Energy management unit (EMU): exchanges information with the other elements of the system and manages the electric resources of users based on an intelligent mechanism. Specifically, this mechanismdefines the schedule of appliances, the operating plan of ESSs and the demand and supply profiles (i.e.,when to buy and inject energy into the grid).

  • Smart grid domains: are the distribution, operation, market, service provider and customers domains of
    the smart grid [2]. A utility company, which is part of the market domain, supplies electric energy to users
    from whom it receives payments according to energy tariffs. Eventually, also a profit-neutral entity, called
    independent system operator (ISO), can be introduced which stands between suppliers and customers.
    The ISO procures electricity from suppliers and sells it to users with the goal of matching supply and
    demand.
All the components of the architecture are connected through a communication infrastructure [3]. Specifically,
one or more Home Area Networks (HANs) are used to interconnect the elements that are within the customer's
domain, while Wide Area Networks (WANs) are required to connect EMUs to the other domains of the smart
grid.





All the components of the architecture are connected through a communication infrastructure [3]. Specifically, one or more Home Area Networks (HANs) are used to interconnect the elements that are within the customer's domain, while Wide Area Networks (WANs) are required to connect EMUs to the other domains of the smart grid.


Demand management mechanisms can be classified according to five different dichotomies:


  • Demand-response (DR) vs. demand-side management (DSM): DR methods are reactive solutions designed to encourage consumers to dynamically change their electricity demand in the short-term, according to signals provided by the grid/utilities such as prices or emergency condition requests. Typically, these techniques are used to reduce the peak demand or to avoid system emergencies such as blackouts. On the other hand, DSM represents a proactive approach aimed at making consumers energy-efficient in the long-term.
  • Individual users vs. community of users: demand management mechanisms can be designed to control electric resource of individual users [4]. However, this approach may have some undesirable effects since it may disturb the natural diversity of consumers, which is fully exploited by the power system to optimize its efficiency in generating and distributing energy. To contain these unwanted side-effects, management mechanisms can control community of users, thus managing their resources based on a system-wide perspective [5]. The natural extension of demand management mechanisms for communities of users is represented by solutions designed for microgrids, which are small-scale versions of electricity systems.
  • Centralized vs. distributed: in case of demand management mechanisms for community of users, two different approaches can be used to control the electric resources of consumers. In centralized solutions, based on optimization methods, all users cooperate in managing their resources and the EMU optimizes a shared utility function [6]. On the other hand, in the case of distributed solutions, users take decisions locally and individually and conflicts among consumers are modeled based on game theoretic frameworks [7].
  • Deterministic vs. stochastic: several parameters of demand management systems, such as renewable energy generation, devices usage and energy prices for future periods, are estimated by prediction methods. In the case of deterministic techniques, these parameters are defined as deterministic data [8], while in the case of stochastic methods, they are represented as random variables in order to consider uncertainty in the decision-making process [9].
  • Day-ahead vs. real-time: demand management mechanisms can use different time scales in controlling the resources of customers. In the day-ahead case, the operating plan of electric resources of users is defined over the next 24-hour time period (or an alternative future time horizon) [6]. To this end, demand management mechanisms require predictions/estimates of some parameters of the system, such as the energy generation of local sources, electricity prices and devices usage preferences for the next day. These data can be defined by learning algorithms executed by the EMU based on information provided by sensors, smart appliances or external sources (e.g., weather forecast). On the other hand, in the real-time case, the users' plan is defined in real-time based on real-time events and data [10].


Despite the great effort in the field of demand management mechanisms, many research issues remain open. First, the majority of existing works are designed to individually and separately manage customers. Cooperative models are clearly more difficult to study and solve since they are typically characterized by time, appliances and customers-coupled constraints. However, in the perspective, they represent the most promising strategy to be implemented in real use-cases. Second, demand management methods are often designed to improve the efficiency of power grids based on proper electric energy tariffs, which implicitly represent the grids needs. However, better performance, from a system-wide perspective, may be achieved by explicitly considering the grids requirements and objectives.


References:

[1] Strbac, Goran. "Demand side management: Benefits and challenges." Energy policy 36.12 (2008): 4419-4426

[2] Locke, Gary, and Patrick D. Gallagher. "NIST framework and roadmap for smart grid interoperability standards, release 1.0." National Institute of Standards and Technology (2010): 33.

[3] Wang, Wenye, Yi Xu, and Mohit Khanna. "A survey on the communication architectures in smart grid." Computer Networks 55.15 (2011): 3604-3629.

[4] Goudarzi, Hadi, Safar Hatami, and Massoud Pedram. "Demand-side load scheduling incentivized by dynamic energy prices." Smart Grid Communications (SmartGridComm), 2011 IEEE International Conference on. IEEE, 2011.

[5] Hatami, Safar, and Massoud Pedram. "Minimizing the electricity bill of cooperative users under a quasi-dynamic pricing model." Smart Grid Communications (SmartGridComm), 2010 First IEEE International Conference on. IEEE, 2010.

[6] Barbato, Antimo, et al. "A framework for home energy management and its experimental validation." Energy Efficiency 7.6 (2014): 1013-1052.

[7] Saad, Walid, et al. "Game-theoretic methods for the smart grid: An overview of microgrid systems, demand-side management, and smart grid communications." Signal Processing Magazine, IEEE 29.5 (2012): 86-105.

[8] Mohsenian-Rad, A-H., and Alberto Leon-Garcia. "Optimal residential load control with price prediction in real-time electricity pricing environments." Smart Grid, IEEE Transactions on 1.2 (2010): 120-133.

[9] Jacomino, Mireille, and Minh Hoang Le. "Robust energy planning in buildings with energy and comfort costs." 4OR 10.1 (2012): 81-103.

[10] Chang, T-H., Mahnoosh Alizadeh, and Anna Scaglione. "Real-time power balancing via decentralized coordinated home energy scheduling." (2013): 1-15.


Can data centers sustain through dynamic electricity pricing markets?

Burak Kantarci, Clarkson University, USA and Hussein T. Mouftah, University of Ottawa, Canada


Since the Environmental Protection Agency’s (EPA) data center report to US congress in 2006 [1], data center sustainability has been a phenomenal topic as data centers have been pointed as the main polluters of the ICTs. Especially, with the advent of cloud computing, corporate data centers are evolving to giant cloud data centers with enhanced computing, storage and communications capability. Indeed, these enhancements are at the expense of increased power consumption which does not only denote the computing power but other non-IT power components such as cooling, uninterrupted power supply, lighting and so on. It can be intuitively said that the lower the computing power gets the lower the non-IT power is. Is it really the gigantic power consumption that makes data centers the major polluters of the cloud dominated era?

As everyone pursuing research, development or service in this subject knows, Greenhouse Gas (GHG) emissions are the main sustainability indicators of any system. Therefore, greening the data centers via energy savings holds if data centers are always powered by the same energy source. Local solutions can be energy-aware workload and virtual machine placement [2], monitoring hot and cool isles in the data center for workload (re)placement [3][4], and sleep scheduling of physical servers. In an inter-data-center network environment, virtual data center (VDC) mapping with the objective of reduced carbon footprint and increased revenue of the provider can be a promising solution for the inter-data center network [5].

Apart from the discussion about consumer behavior, as a part of demand side management, dynamic electricity pricing has potential in integrating renewable resources into the electricity grid. Thus, electricity prices can be lowered when the renewable energy supply is higher and vice versa [6]. It can be hypothesized that data centers can take advantage of dynamic electricity pricing if they can migrate workloads among themselves. Such a system will require demand profile prediction by a centralized information service which communicates with network resource managers, smart grid communication network, resource coordinators and computing resource managers.

We have recently proposed a framework which enables workload migration among data centers based on time varying electricity prices at different locations [7]. A minimalist visualization of the proposed system mainly consists of the inter-data center network and the Smart Grid Communication Network (SGCN). The inter-data center network consist of an IP over WDM transport network and the data centers associated with the backbone nodes, and the SGCN consist of the power generators, substations, transformers, smart meters and the control center [8]. SGCN components are interconnected via aggregation switches which are connected to the inter-data center network via a metro ring network. Based on the changes in the demand profile, the inter-data center network is virtualized periodically in order to achieve the design goals, namely energy-efficiency and minimum electricity costs for the network and the data center operators.

Our research has reported that such a system is promising in terms of electricity bill costs for the data center operators, as well as the network operators when ToU tariffs are adopted. Moreover, energy consumption is also reduced if each data center is obliged to keep certain amount of its original workload prior to migration, which is further promising in terms of data center sustainability. Further, having shown that path delays for workload migration as well as service provisioning can be limited if ToU-awareness is integrated into the information service which is expected to virtualize the network and determine the workload migration map.

Despite the advantages of dynamic electricity price-aware design of inter-data center networks, the following points remain as challenging open issues for the researchers in this field:

  • Determining the workload type(s) to be migrated: As research reports that each data center has to host a certain amount of the workload prior to migration, some workload types can be identified in advance and enforced to remain in the original data center. Service Level Agreements play critical role in the decision of workload types to be migrated.
  • Determining the workload migration time dynamically: In our recent research, we have let the information service periodically calculate a workload migration map and reconfigure the virtualized topology of the inter-data center network. Based on the dynamics of varying electricity prices at different locations, workload migration, as well as topology reconfiguration can be triggered adaptively to ensure further sustainability of the data centers.

  • Minimum QoS degradation: Workload migration and topology reconfiguration should ensure minimum QoS degradation. Thus, service disruption for particular workload types should strictly be avoided. The easiest method to avoid service disruption is to avoid migration of the workload of higher priority service to another data center. On the other hand, corresponding workload’s remaining in the original data center increases the electric bills as the workload is not taking advantage of time varying electric prices. Besides, this situation may degrade sustainability since the workload is not migrated towards the locations where renewable resources are supposed to be available. Trade-off mechanisms should be adopted to address this conflict.

  • The impact of the several time varying electricity pricing policies: Although the entire discussion focuses on TOU pricing, real time pricing (RTP) and critical peak pricing (CPP) are also possible dynamic pricing policies however, the consumers are more reluctant to switch to the tariffs where electricity prices vary more frequently. On the other hand, assuming that dynamic electricity prices have the potential to reflect the availability of renewable resources, data centers can further benefit from other highly dynamic pricing policies such as RTP and CPP. Therefore, researchers are encouraged to focus on consumer behavior models to help RTP and/or CPP-based solutions become comfortably adopted.

  • Location policies for new data centers: The crucial questions for data center operator are the following: Where to build the data center? Here are some viable strategies: “Build the data center where the demand is more intense”, “Build the data center where more renewables are available,” or “Build the data center where electricity prices are lowest (assuming flat rates are applied)”. Data center location problem has been widely studied [9], and all of these strategies have pros and cons in several aspects. If the dynamic electricity pricing policies can be considered as indicators of renewable energy; the data center operators can minimize their operational expenses via long term analysis of varying electricity prices, which in turn leads to less GHG emissions and more sustainable environment.

Once all of these points have been addressed, it will be possible to answer the question in the title by confirming that data centers can sustain through dynamic electricity prices. It is worthwhile noting that the energy consumption of the data centers in the US is still at the order of 2% of the nationwide energy consumption as reported by EPA by the end of 2011. Therefore, all of the points mentioned above introduce emergent challenges to be addressed by the industry and academia towards a sustainable environment.

References:


[1] Environmental Protection Agency (EPA) “Report to Congress on server and data center energy efficiency” [Online] http://www.energystar.gov/ia/partners/prod_development/downloads/EPA_Datacenter_Report_Congress_Final1.pdf (2007).

[2] A. Beloglazov, J. Abawajy, and R. Buyya, “Energy-aware resource allocation heuristics for efficient management of data centers for cloud computing,” Future Generation Computer Systems, vol. 28, no. 5, pp. 755 – 768, 2012.

[3] L. Parolini, B. Sinopoli, B.H. Krogh, and Z. Wang, “A Cyber-Physical Systems Approach to Data Center Modeling and Control for Energy Efficiency,” Proc. of the IEEE, vol. 100/1, pp. 254 –268, Jan. 2012.

[4] H. Viswanathan, E. K. Lee, and D. Pompili, “Self-organizing sensing infrastructure for autonomic management of green datacenters,” IEEE Network, vol. 25/4, pp. 34 –40, 2011.

[5] A. Amokrane, Mohamed F. Zhani, R. Langar, R. Boutaba, and G. Pujolle, “Greenhead: Virtual data center embedding across distributed infrastructures,” Cloud Computing, IEEE Transactions on, vol. 1, no. 1, pp. 36–49, 2013.

Greening the cloud using renewable-energy-aware service migration

U Mandal, MF Habib, S Zhang, M Tornatore, B Mukherjee IEEE Network Vol. 27, no. 6, pp. 36-43, 2013

 
Modern cloud computing systems require a large amount of computing, storage, and networking resources. Datacenters are used to provide the computing and storage resource pool, while wide-area networks interconnecting datacenters provide the communication media among cloud services and consumers. As of today, it has been widely acknowledged that both communication networks and datacenter infrastructures consume substantial amount of energy, which in turn makes them a significant contributor of carbon and green-house gas emissions.

Traditionally, a two-pronged approach is being taken to overcome this problem. First, reducing the energy consumption of cloud-computing infrastructure by increasing energy-efficiency in different parts of the system, and second, increasing the consumption of renewable or green energy in place of fossil-fuel based brown energy in datacenters and networks.

Focusing on the renewable energies, their adoption in could computing has relevant challenges – renewable energy sources are volatile and intermittent, and are often characterized by temporal and geographical disparity between energy demand and the production of green energy.

The main idea of the proposed study is that cloud-computing systems can be an excellent vehicle to reduce this disparity of demand and supply by exploiting its unique characteristics of energy-demand relocation through virtual machine (VM) migration among datacenters.

Physical servers in datacenters are virtualized as VMs. Among other benefits of virtualization, a useful feature of virtualization is the possibility to migrate VMs across physical servers. The study investigates how to perform renewable-energy-aware VM migration in order to relocate energy demand using dynamic and flexible cloud-resource-allocation techniques. First, the article describes how VM migration techniques over a wide area network can be used to relocate energy-demand. Then it proposes and analyzes algorithms to perform renewable-energy-aware service migration. Experimental results from a US-wide network show that up to 30% of total brown energy consumption can be replaced by green energy with only an increase of as low as 2% of total energy. Note that, small increase of energy is inevitable while using VM migration (energy consumed by the migration process itself), and this increase is quite negligible, especially considering the substantial replacement of brown energy with green energy.

While renewable-energy-aware service migration is an excellent choice to reduce carbon emission and for sustainable future, more avenues must be explored to exploit renewable energy to its full potential. While a single optimum bandwidth for the entire migration duration is key, multiple bandwidth can also be considered. Also temporal migration of delay-tolerant workloads, in addition to spatial migration, is an important research problem. Jobs could be scheduled for future when it is predicted to have more RES available. Finally, exploiting VM migration for content delivery services is another important aspect of the problem that needs to be explored..


Uttam Mandal (umandal@ucdavis.edu) received his B.S. degree in computer science and engineering from Javadpur University, Kolkata, India, in 2005. He completed his M.S. degree in computer science from University of California, Davis, in 2010, and Ph.D. from the same university in June, 2014. His research activities focused in the fields of energy and cost-effective cloud infrastructures, content delivery networks, and data center network design and operation.


Conference Showcase

Lorenz Hilty, Mattias Höjer, Patricia Lago and Josefin Wangel  

    
ICT is a transformational technology bearing both opportunities and risks for sustainable development. The "ICT for Sustainability" (ICT4S) conferences areabout utilizing the transformational power of ICT for making our world more sustainable: saving energy and material resources by creating more value from less physical input, increasing quality of life for ever more people without compromising future generations´ ability to meet their needs. While the potential of ICTs´ contribution to sustainable development has been increasingly recognized during the last decades, implementing this potential has proven to be a challenge. One of the main obstacles is the cooperation between the academic disciplines, such as engineering sciences and social sciences, but also between academia, industry, and politics. From its beginning, the conference therefore had a strong focus on exchanging ideas and mediating between different intellectual cultures. 

The first ICT4S conference took place at ETH Zurich in February 2013 with over 200 participants from 50 countries and all continents. In addition to the proce
edings [1], selected papers were later published in a thematic issue of “Environmental Modelling and Software” [2] and Book in the Springer series "Advances in Intelligent Systems and Computing" [3]. The proceedings contain in the appendix a set of recommendations to stakeholders on "How to improve the contribution of ICT to sustainability" [1] (pp. 291-295), which were endorsed by the participants. 


Although not originally planned by the organizers of the first conference, the conference became a series: ICT4S 2014 was held in Stockholm in August and was organized by KTH Royal Institute of Technology in cooperation with VU University Amsterdam. For the second conference, the peer-review process was further sharpened, leading to a final result of 49 (50%) accepted papers based on full-paper reviews. The papers were published with Atlantis Press and open access [4]. ICT4S 2014 attracted 150 people from 34 countries to the main conference and 200 people including workshops and study tours. Taking the original ideas on increased collaboration from the first conference one step further, a lot of efforts were put into the conference outline. The most concrete example of this was the innovation “ConverStations” where research papers were presented in small groups and repeated three times, giving much more opportunities for interaction than is common at research conferences. Altogether the response from participants was overwhelmingly positive, and it got clear that an ICT4S-community has been built.

The success of the first two conferences is now handed over to the organization of ICT4S 2015. That conference is in preparation (jointly with EnviroInfo 2015) and will be organized by the University of Copenhagen in cooperation with the European Environment Agency (EEA) and Bristol University [5] in September 7-9 2015.



About ICT4S-conferences:

The ICT4S-conferences are held together by a standing committee responsible for long-term development and for appointing venue and general chair for upcoming conferences, ict4s.org.


References

[1] ICT for Sustainability: Proceedings of the First International Conference on Information and Communication Technologies for Sustainability. E-Collection ETH Institutional Repository (2013) http://dx.doi.org/10.3929/ethz-a-007337628

[2] Modelling and Evaluating the Sustainability of Smart Solutions (Thematic Issue), Environmental Modelling and Software 56 (2014) http://www.sciencedirect.com/science/journal/13648152/56

[3] Hilty, L.M.; Aebischer, B. (eds.): ICT Innovations for Sustainability. Advances in Intelligent Systems and Computing 310. Springer International Publishing (2015) http://link.springer.com/book/10.1007%2F978-3-319-09228-7

[4] Höjer, M., Lago, P., and Wangel J. (eds): Proceedings of the 2014 conference ICT for Sustainability, Atlantis Press (2014)

http://www.atlantis-press.com/php/pub.php?publication=ict4s-14

[5] EnviroInfo and ICT4S Conference 2015: 2015.ict4s.org