- Energy efficiency is one of the very few measures widely used for evaluating green and sustainable software systems. This paper investigates the current measures of software sustainability from the four different software sustainability perceptions: Sustainability in Software Development, Green Software Systems, Software for Sustainability, Sustainability of the Software Eco System and Software Sustainability Dimensions (Economic, Social, Individual, Technical and Environment). While exploring the literature on green and sustainable software systems, measures of green software and software sustainability were identified, compiled and classified according to the four sustainability perceptions.
Sustainability is now one of the world major challenge
[
This research explores the ongoing perceptions in the software engineering domain with the goal to identify the current and future measures used in the evaluation of green and sustainable software. Via triangulation of data from diverse sources, the measures are clustered into the four perceptions of sustainability in software engineering and sustainability dimensions. The long term goal of this research is to answer the following challenging questions: what are the current measures used in evaluating green and sustainability aspects of software systems and how can these measures be grounded in the software sustainability measurement theory.
Furthermore a checklist and guide approach that demonstrates how to include the objective of environmental sustainability from the very early steps of software development can assist in identifying key stakeholders. This will facilitate the ability to accommodate new objectives of improving the environmental sustainability of software systems [13]. Roher et al. [14] suggests the use of sustainability requirement patterns (SRPs), which will provide software engineers with guidance on how to write specific types of sustainability requirements with the goal to overcome the barriers of incorporating environmental sustainability into the requirements engineering process.
Green software is an environmentally friendly software that consumes less energy, provides less impacts on environment and support carbon management [15]. It is also software that fulfils high level requirements, ensuring the software engineering process, maintenance, and disposal saves and/or reduces resource waste [16] [17]. Green software is divided into four parts: software that is energy efficient during execution, software that are embedded to execute and support smart operations in green manner, software to produce environment viable products and policies [18]. The goal of green software engineering is to provide supports for efficient consumption of natural resources while continuously monitoring, evaluating and optimizing the aftermath effects caused during the software system life cycle [19].
Erdélyi [20] paper provides an overview of different activities and advice on what to do in order to develop green software which uses energy efficiently and produce less waste. The paper highlights three ways software engineering can be green such as: produce green software, produce software to support environmentally consciousness (green by software) and produce less waste during development.
Dick et al. [21] provides basis for the right way to engineer green software systems using development process that ensures that the positive and negative effects of the software is continuously monitored and evaluated in order to optimize the software over its life cycle to be more green (environmental friendly).
Colmant et al. [22] presented researches on to improve the software-energy efficiency on multi-core systems. Colmant et al. [28] motivations were driven by the huge impact of the ICT on the world CO2 emissions which represents 2%. Calero et al. [4] highlights some of the meanings of green software notably a software that consumes less energy to run and produces as little waste as possible during its development and operation. Largely, research on green software has focused more on energy consumption and environmentally friendly software systems.
There has been some interest in various domains such as manufacturing, energy sector, transportation and for different application in recycling, product packaging, data center setup, gas emissions. Some of the good examples are in grid computing, in Human Computer Interaction (HCI) to change the habit of people.
In [23], authors presented a software system that support sustainable lifestyles with an example of a domestic plant guild to show how sustainable human systems can effectively support a sustainable lifestyle, which can reduce the cost of living as well as the ecological footprint. Penzenstadler et al. [24] highlights vision for systems that will be supporting sustainability in the future (2029) with a set of fictional abstracts around the concepts of sustainability, complexity, collapse, and resilience of ICT systems.
Software can also provide support for sustainability in different domains such as: x The use of software systems for tracking gas emissions x Software for climate and disaster prediction x Smart infrastructural management software x Enterprise carbon and energy management software x Smart transportation software to reduce CO2 emissions.
Sustainability Knowledge and Learning Management software Software for environmental awareness on wildlife and plants
Today software systems are the pillars of the economy, the software eco system is probably the biggest system in the world we human created. Software eco system has been defined according to Jansen et al. [25] as a set of actors functioning as a unit and interacting with a shared market for software and services, together with the relationships among them. These relationships are frequently underpinned by a common technological platform or market and they operate through the exchange of information, resources and artifacts.
Thus, the sustainability of software ecosystem involves the sustainment of the global system of software systems and services covering aspect of how different sub systems form a huge interconnected system and all the interactions. It covers all different components such as hardware, software and network that is used to resolve complex relationships among companies/organizations in all the different sectors and industries [26].
Sustainability of the software ecosystem entails how can the system of software systems endure with the evolving user requirements and usage overtime with less negative impact on the environments, social, technical and humans. This means the ability of software ecosystem to continue to function and evolve irrespective of any glitch is some part of the ecosystem and should continuously fulfil users’ needs.
III. PERCEPTIONS OF SUSTAINABILITY IN/FOR SOFTWARE
SYSTEMS
In this research, we defined sustainability as a quality construct in the same ways other factors are defined (see, for example, the ISO 25 000 family of standards). In our perception sustainability aims to create balance in the way humans live, produce, and use products and services (resources) with the objective to have less negative impact on the environment and promote the wellbeing of all living species. This means the capacity of software systems to endure in certain ecosystems under current and future conditions while satisfying the needs of users today and tomorrow with minimum negative impact on the environment; at the same time supporting business growth and societal values.
Currently, the dimensions of software sustainability are known and classified into five: economic, environment, social, individual and technical [27] but there is currently no clear categorisation for the perceptions of sustainability in/for software engineering. This section explains the categorization of software sustainability perceptions based on the literature review from the background section. Software sustainability evolution today can be perceive from one of the following perception (see Figure 1); Sustainability in Software Development, Software for Sustainability, Green Software Systems, Sustainability of Software Ecosystems. x
Sustainability in software development (Development): this refers to the processes involve in the development of software (software development life cycle). x x x
effect): This is the total impact of the entire software ecosystem (systems of system)
The advancement of software sustainability from the four perceptions has received different level of research attention and contributions. Sustainability in software development, Green Software system we observed has the most important advancement in research compared to software for sustainability and sustainability of software ecosystem that were not fully explored.
Figure 1 portrays the categorization of software sustainability perceptions.
SUSTAINABILITY IN/FOR SOFTWARE SYSTEMS
This section presents different research work relating to green and sustainable software system measures and measurement. According to Britannica [28], measurement is the science of assigning of a quantity, either quantitative or qualitative, to a characteristic of an object or event, while making it comparable to other objects or events. Here object is the software and event is the development process.
Sustainability measurement is still a new idea [29] [30] [31] [32]. Indeed, Lami et al. [31] stated that there are few studies about ‘what’ aspects of sustainability to measure and ‘how’ to do it. Calero et al. [33] highlighted that nowadays, sustainability is a key factor that should be considered in the software quality models, though there has less research channelled towards sustainability measurement. Seacord et al. [29] stated that planning and management of software sustainment is impaired by a lack of consistently applied, practical measures, and there is no central theoretical framework on measurement of software sustainability.
One of the most referenced sustainability measurement model for software system is the GREENSOFT Model [7]. It is a conceptual reference model for “Green and Sustainable Software”, which has the objective to support software developers, administrators, and software users in creating, maintaining, and using software in a more sustainable way [34]. Another software sustainability measurement approach is the Sustainable Business Goal Question Metric (S-BGQM) [35]. It encourages the incorporation and measurement of sustainability during the entire software system development processes. Kramer [36] also wrote about sustainability measurement by proposing some set of questions that should be addressed by any sustainability framework.
A study for monitoring software energy hotspot proposed power model for software energy cost formula as Esoftware = Ecomp +Ecom +Einfra, where Ecomp is the computational cost (i.e., CPU process- ing, memory access, I/O operations), Ecom is the cost of exchanging data over the network, and Einfra is the addi- tional cost incurred by the OS and runtime platform (e.g., Java VM) [37]. The study focused on energy consumption of CPU and network demanding software at different levels of granularity. Also, the formula proposed for software energy efficiency (EF) is UsefulWorkDone/UserdEnergy [38]. This generic measure provide a way for evaluating the energy consumption of different software parts and modules using white box testing to measure which parts are consuming more energy and to see which parts can be optimized for efficient energy usage.
In order to facilitate research on energy usage attribution, software energy footprint lab was setup to provide insight on energy footprint measurements with results interpreting hardware dissipation profiles for various servers under different kinds of software stress [39]. This shows the relations between different hardware resource and the amount of resource required by the running software in relation to the power consumption.
Furthermore, a support tool is presented to analyze legacy systems in order to estimate the energy consumption and detect parts of the system with higher energy consumption. Using the profiling technique, the tool instrument legacy Java systems in order to keep track of its execution. This information, together with the energy consumption, enables the engineer to analyze legacy system consumption detecting energy peaks in the system [40].
Additionally, a modular Energy-Aware Computing Framework (EACOF) is proposed as a way to allow access to energy consumption information of software through API calls. The EACOF is separated into two task for collection and utilization of dynamic energy consumption data which reduce development and maintenance overhead required for the successful completion of each task[41]. Another approach is also proposed for monitoring power consumption of software in order to assist software designers and developer to reduce software power consumption and have better energy efficiency [42]. This approach currently monitors power consumption at source code level, this approach will provide better insights on software energy consumption if extended to the hardware running the software.
As summarized in Table 1 and the research work detailed in [43] [44] [45] and [46], other measures of green and sustainable software have been on software and hardware energy consumption with less research for measures covering software sustainability dimensions such as individual, social, economic and software sustainability perceptions (Software for sustainability and Software ecosystem).
The measures detailed in Table 1 are structure based on categorization of software sustainability and green measures for software sustainability dimensions and the four sustainability perceptions. Each column after the main title has a “YES or No” to indicate if the proposed measure in the research paper cover any of the categories listed in Table 1.
Most of the measures descriptions does not explicitly indicate that the authors considered sustainability dimensions. Base on the descriptions and explanations of the authors for all measures, we have categorized those measures according to the right sustainability dimension (Economic, Social, Individual, Technical and Environment) to show how it relates to the four sustainability perceptions.
MEASURES FOR GREEN AND SUSTAINABLE SOFTWARE LINKED TO SUSTAINABILITY DIMENSIONS AND PERCEPTIONS
Table 1 provides details of measures attributed to green and sustainable software. From Table 1, it can be identified that most measures focused on energy efficiency or power consumptions. With most focus on green software, there is a limitation on having a holistic approach towards software sustainability measurement. The measures of software sustainability should consider the following: x Human (End users) system interaction: involves the measures of the system sustainability based on how it impacts on users and their level of awareness about sustainability and green. It entails the well-being of the software users’ community and the changing of the human mindset. x Software system developers: evaluate the sustainability of the processes and practices for the development and integration of sustainability in software systems.
One of the key question/concern that should be clearly answered by a sustainability measurement framework is the difference between the different scales of software measurement and the interpretation of these scales of measurement for sustainability. The problem of software sustainability measurement is not only in measuring but Yes
No
No
No rather giving meaningful interpretation of what the measurement means. For example today, fridges are categorized using A+, A++ and A+++ for quantifying and measuring its energy efficiency. Normally A+ consumes less energy, A++ has better energy efficiency than A+ and A++ has the best energy efficiency in today market. According to the EU Directive 92/75/EC which established an energy consumption labelling scheme [52], there are different descriptions of the measures that quantify why Fridge is labelled A+, A++ or A++ based on its energy consumption.
In the same line, there is need for a foundation or framework to ground the different measures for software sustainability measures and measurement with clear interpretation.
Currently, there is not enough firm scientific basis for important choices on how sustainability related factors should be defined and measured, the varying purposes for which the measures are used. This makes it difficult to effectively and efficiently evaluate software sustainability using the right measures.
In this position paper, we summarized the research
results on the categorization of software sustainability
perceptions. Using the identified four perceptions of software
sustainability, we referenced the current measures to each of
the four perceptions. The major focus of all identified green
and sustainable software measures are on green software.
Energy efficiency has received the most attention. Research
work is needed to identify and assess the validity of other
measures related to the other perceptions. Research on
measures of sustainability has to be grounded in the tradition
and theory of software measurement. This requires
considering software sustainability as a quality attribute and
define it in the same way other attributes are defined.
[