ASCETiC: Adapting Service lifeCycle towards EfficienT Clouds

Reducing energy consumption is increasingly gaining attention in the area of Cloud computing, as means to reduce costs and improve corporate sustainability image. ASCETiC is focused on providing novel methods and tools to support software developers to optimise energy efficiency and minimise the carbon footprint resulting from developing, deploying and running software in Clouds. At the same time, quality of service, experience and perception are still taken into account, so energy efficiency will complement them and boost Cloud efficiency at several dimensions. ASCETiC primary focus is to relate software design and energy use, which will depend on the deployment conditions and the correct operation of the software by means of an adaptive environment. This paper presents specific objectives for the project, as well as requirements, business goals and architecture for the resultant Open Source Cloud stack providing energy efficiency at software, platform and infrastructure Cloud layers. 1 The ASCETiC Project Information and Communication Technology (ICT) constitutes a diverse array of economic, industrial and social activities. Energy efficiency is increasingly important for its future. The increased usage of ICT, together with growing energy costs and the need to reduce greenhouse gases emissions call for energy-efficient technologies that decrease the overall energy consumption of computation, storage and communications. Forecast for ICT [1] indicate a steady growth of carbon footprint (CO2) of about 4 percent per year, accounting for a total of around 70 percent growth between 2007 and 2020, when the estimated total carbon footprint for the ICT sector will be about 1,100 million tons in 2020. Enhancing the energy efficiency of ICT is therefore important, not only to reduce power consumption and CO2 emissions, but also to stimulate the development of a large leading-edge market for ICT-enabled energy-efficient technologies that will foster the competitiveness of the industry, improved Return on Investments (ROIs) and result in new business opportunities. Google, the most-used Internet service, provides a very good example for understanding energy usage and carbon footprint of Cloud services. It is estimated that Google manages over one million servers and processes one billion search request daily. The operation, production and distribution of these servers produce huge amounts of CO2, assessed, depending on the source, of a value between 0.2g [2] and 1g [3] of CO2 per search, amounting to a daily total of one thousand tons of CO2. This is just one example, but it offers an order of magnitude to assess the ecological impact of day-by-day ICT activities. Traditionally Green IT research and development have focused on energy efficiency for hardware and data centre facilities. Energy efficiency for software has only been marginally considered, mostly in mobile computing, aiming to optimize the duration of devices’ batteries. Although it has a direct impact on system’s energy consumption; software usually controls how computing is utilized. Covering the full service lifecycle, from application design, development, deployment and operation, it is crucial to determine and optimize the energy usage of the complete system, considering software and hardware as interrelated mechanisms. Cloud computing presents a model in which IT infrastructure is leased and used according to the need of the enterprise. The benefit of this model is that it converts capital expenditure of an enterprise into operational expenditure. Although building, deploying and operating applications on a Cloud can help to achieve speed, scalability and maintain a flexible infrastructure, it brings about a variety of challenges due to its massive scalability, complexity, as well as dynamic and evolving environments. ASCETiC is concerned with the topical issue of energy efficient computing, specifically focusing on design, construction, deployment and operation of Cloud services. It argues that research is needed to propose novel methods and develop tools to support software developers in monitoring, minimizing the carbon footprint and optimizing energy efficiency resulting from developing and deploying software in Cloud environments. Such research addresses the need for continued development of infrastructure support for Clouds in order to optimize, monitor and reduce carbon footprint and costs for Cloud providers and end users. The major contribution to the carbon footprint of IT software in general is energy consumed in its operation, thus the primary aim of ASCETiC is to relate software design and energy consumption. Although energy use is of relevance across all software design and implementation, ASCETiC makes specific reference to Cloud-based service operations: the emergence of Cloud computing with its emphasis on shared software components which are likely to be used and reused many times in many different applications makes it imperative that the software to be developed is as energy efficient as it possibly can be. Therefore, ASCETiC primary goal is to characterize the factors which affect energy efficiency in software development, deployment and operation. The approach focuses firstly on the identification of the missing functionalities to support energy efficiency across all cloud layers, and secondly on the definition and integration of explicit measures of energy requirements into the design and development process for software to be executed on a Cloud platform. ASCETiC will measure how software systems actually use Cloud resources, with the goal of optimizing consumption of these resources. In this way, the awareness of the amount of energy needed by software will help in learning how to target software optimisation where it provides the greatest energy returns. To do so, all three layers in Cloud computing: Software, Platform and Infrastructure will implement a MAPE (Monitor, Analyse, Plan and Execute) loop. Each layer monitors relevant energy efficiency status information locally and shares this with the other layers, assesses its current energy status and forecasts future energy consumption as needed. Actions can then be decided and executed according to this assessment. 1.1 Vision and Approach Basically, the ASCETiC approach focuses firstly on the identification of the missing functionalities to support energy efficiency across all Cloud layers, and secondly on the definition and integration of explicit measures of energy and ecological requirements into the design and development process for software which can be executed on a Cloud platform. ASCETiC’s goal is to characterise the factors which affect energy efficiency in software development, deployment and operation. Our main novel contribution is the incorporation of a novel approach that combines energy-awareness related to Cloud environments with the principles of requirements engineering and design modelling for self-adaptive software-intensive systems. Therefore, the objectives of ASCETiC are: Objective 1: To extend existing development models for green software design, supporting sustainability at all stages of software development and execution. Objective 2: To develop and evaluate a framework with identified energy efficiency parameters and metrics for Cloud services. Objective 3: Develop methods for measuring, analysing and evaluating energy use in software development and execution. Objective 4: To integrate energy efficiency into service construction, deployment and operation leading to an Energy Efficiency Embedded Service Lifecycle. 2 Requirements Gathering and Elicitation To achieve the vision of an efficient Cloud operation, the ASCETiC project starts with specifying the business and technical requirements for developing appropriate methods and tools matching with Industry needs. To analyse business and technical requirements in parallel, two complementary approaches are followed. For initiating business requirements gathering, a market analysis [4] sets up the stage. Besides, interviews with the two Industry ASCETiC partners, namely ATC and GreenPrefab, as well as other 15 interviews with Industry members beyond the ASCETiC consortium help to derive a set of general business goals to achieve when building the ASCETiC solution. For eliciting technical requirements, ASCETiC partners initially performed a thorough review of the state of the art [5] which provided some useful insight towards ASCETiC architecture. From the state of the art review and initial architecture reflection, the goaloriented requirement engineering method KAOS (Knowledge Acquisition in Automated Specification) [7] is performed to hierarchically specify technical goals and finally elicit technical requirements. 2.1 Requirements Analysis Capturing energy related requirements is challenging because in standard requirements classifications such as the ISO-9126/ISO25030, there is little room for energy requirements except some weak resource-related notion with no reasoning capability about energy requirements and potential interdependencies or even conflicts with other kinds of non-functional requirements like security, availability, time-responsiveness... Actually most of the time energy requirements are not captured as stressed by [6] and there is a need for better elicitation process to capture energy issues such as the available power budget requirements engineers can work with, how to reason about energy requirements across the design space or event at runtime. This can be achieved using a goal-oriented requirements engineering methodology such as KAOS [7] Explicitly capturing and reasoning about energy requirements is key to have a better understanding of the impact in terms of costs. Currently, the billing of an ICT service is hardly connected to its "real" energy cost and thus lacks transparency [8]. The ASCETIC vision of how to deal with energy requirements together with other types or requirements and how to manage them is divided in three stages:  At requirements time: explicitly dealing with energy goals as described above. Organising them into a reusable pattern library is also useful.  At design time: identifying relevant energy-related feature of the target architecture and possible variability points which can be used to explore the designed space. Practical means such as UML annotation can be used to support this activity and relate them with energy requirements.  At run-time: translating requirements into a Service Level Agreement (SLA) (i.e. possibly “green SLA” for energy requirements) with associated Key Performance Indicator (KPI). A dynamic architecture can dynamically enforce those SLA based on a "Collect-Analyze-Decide-Act" self-adaptability control loop as described in [9]. Important research questions that will be addressed by ASCETiC are the normalisation of energy measurements, the mapping between hardware, VM and software level, the management of KPIs of contributing/conflicting goals as well as the identification of variability points available for (self)-adaptation. 2.2 Stakeholders and Questionnaires main findings To propose a solution appropriate to the Cloud market, ASCETIC collected business requirements by consulting staff and management of various companies covering most pertinent roles identified in the previous section. This process has been performed by conducting 17 interviews with relevant stakeholders that constitute a representative sample of Cloud professionals today, what includes different roles such as Cloud service providers, Cloud users and a Cloud business analyst. Except for roles from Intermediaries, such as Deployer and Auditor, where the sample of interviewees is minimal, for all other roles, the number of interviewees is adequate for gathering enough information to steer ASCETIC in an appropriate direction. Cloud Customers expectations are of special interest at this stage, as having a clear understanding of their motivations and views will help craft an ASCETIC solution that align with this market needs. 2.2.1 Questionnaires main findings ASCETiC motivation is built on top of EU’s Europe 2020 Strategy for smart, sustainable and inclusive growth and of the transition to a resource efficient economy. It was therefore important to get insights about sample interviewees’ awareness of this initiative. In this regard, results are positive, 70% (12) have previous knowledge on the initiative, of course at different levels depending on the profile of the interviewee. In research organizations, 100% of the interviewees are aware of Europe 2020 Strategy, 60% in commercial organizations and 50% in public organizations. 2.2.1.1 Cloud Business Customer Role 90% of interviewees are interested in services consuming less energy. However the decision factor and business driver for Cloud service selection is cost. The capability to assess energy consumption by means of real time monitoring mechanisms and measurements is perceived as an interesting feature, however there has to be a balance among ecological and economical sustainability. 56% of the interviewees would choose a similar eco-service at the same cost, and the average cost reduction to consider it would be 7%. Only 50% of the interviewees perceive energy consumption as a problem from a customer perspective. When negative responses occurred, emphasis was put on highlighting the fact that it is a cost factor for providers not for consumers of the services. Those ones considering it a problem report that the main drivers for consumption of ecoefficient services are corporate social responsibility, cost and eco-sensitivity. The majority of the interviewees (nearly 60%) consider that new pricing schemes will charge users based on their actual energy consumption are fair. However, interviewees make an important remark: users have to be able to properly control and monitor both resource usage and energy metrics. Responses show that cloud customers (65%) will be willing to sacrifice certain QoS parameters such as performance, security or availability but only by their specific selection and for certain services that are not mission critical. For all the rest, QoS agreed in the SLA have to be of mandatory fulfilment for the provider and performed optimisation to keep agreed service terms. Penalties and Rewards are not considered of much interest by Cloud Business Customers, less than 20% considers them as a potential motivation. The other motivations that make customers consider a service, in addition to cost, are: use of Green or Brown energy sources (75%) and image (33%), in addition to cost. 2.2.1.2 Service Business Operator Operator roles that consider energy consumption an important problem or a very important problem are approximately 57% of the interviewees and only two of the ones that are responsible in their organization of the energy bill don’t consider it an important topic. Motivations for energy concerns are clearly cost and to a lower extent corporate social responsibility, emissions and organization image. Service providers’ interviewees (71%) understand that offerings that consider energy consumption similarly to other quality of services aspects could be beneficial for their business. Service providers interviewees find that the most effective action that relevant authorities could implement to encourage them to propose lower energy consumption or emissions services would be Tax Rebates. Eco-labels are perceived as interesting but clearly to a lesser extent. Interestingly some of the interviewees think that adoption of greener services will happen independently of policies, although they represent less than 10% of the interviewees. Displaying energy behaviour is perceived in general as positive by 62% of the service providers interviewed however it is clear from the responses that there is not yet a market push in the IT/Cloud sector. Respondents associate this to a reduced target of their customer base and a need for cheaper services. From the providers’ perspective the incentive that is found more interesting to grant customers in order to promote energy responsible behaviour is to provide Eco-labels to customers. The use of flexible eco-aware pricing models as an incentive is perceived of less interest. Based on this, it seems the providers’ interest is more aligned with internal cost reduction than in making their customers being part of the process.