A Service Computing Manifesto: The Next 10 Years

Current it advances can be likened in their effect and impact to the deep economic and societal transformations that were brought about by the industrial revolution. Recent years have seen the expansion of services in all sectors of the economy, building on the expansion of the Internet.

Back to Top

We define service computing (alternatively termed service-oriented computing) as the discipline that seeks to develop computational abstractions, architectures, techniques, and tools to support services broadly. A service orientation seeks to transform physical, hardware and software assets into a paradigm in which users and assets establish on-demand interactions, binding resources and operations, providing an abstraction layer that shifts the focus from infrastructure and operations to services.

But, as we argue here, service computing has not fully reached its potential. Technological advances provide an increasing opportunity for service computing. To avoid the mistakes of the past, we propose a manifesto^a—a community declaration of objectives and approaches—as a way to establish common ground among researchers in the field and to guide its future development.³

This manifesto first takes stock of the current state of service computing and then maps out a strategy for leveraging emerging concepts and technologies to deliver on the full potential of the service paradigm. It identifies the major obstacles that hinder the development and potential realization of service computing in the real world; proposes research directions; and draws a roadmap to enable the service computing field to redefine itself and become one of the powerful engines for social and economic activities.

Evolution of IT: Services as the next layer of the computing value chain. Computing has progressed dramatically over the past few decades in delivering automated solutions for an increasing number of areas. In what follows, we provide a historical perspective on the evolution of computing. In the early days of computing, the challenge was to represent information in a machine-readable format that consisted of bits and bytes, called data. Over time, there was keen interest in complementing data with meaning, thus transforming it to information. With further advances in computing came the idea of adding reasoning to information, thus giving rise to knowledge.⁶ The need for higher levels of abstraction has recently led to the notion of adding action to knowledge, resulting in services. Therefore, the abstract definition of a service comes to fulfill the need to act and deliver on knowledge, that is, to provide a way for knowledge to be useful. Accordingly, services are de facto presently considered as the highest level in the computing value chain^b (see Figure 1).

Services are ubiquitous in today’s social and economic environment. Examples of such services include healthcare, financial management, human resources, and tourism planning. What distinguishes services from other computing paradigms is their ability to work in a competitive environment where the key parameter to distinguish between similar services is their quality. Knowledge in itself is not sufficient, but needs to be acted upon to bring about benefits. In the case of services, it is the ability to use quality of service (QoS) as a key discriminant to choose between services that provide the “action” on knowledge about services.^7,19,24

As economies undergo significant structural changes, digital strategies and innovation must provide industries with tools to create a competitive edge and build more value into their services.⁴ Advances in online service technologies are transforming the Internet into a global workplace, a social forum, a means to manage one’s individual affairs and promote collaboration, and a business platform for service delivery. Additionally, organizations are competitively compelled to provide service interfaces to their online services, allowing third-party developers to write auxiliary or satellite “apps” that add new uses to the original service, enrich its features and accessibility, and enhance its agility.

Web service technologies have been developed somewhat independently from the notion of service. They have been at the center of intense research in the past 15 years. In the enterprise market, sales of ready-made software or hardware products are rapidly being replaced by the provisioning of customized IT Web services, aimed at individual problem solving. Service computing takes a broader view and has emerged as a cross-disciplinary research field that studies the science and technology underlying the popularity of the IT service industry. The ultimate goal of service computing is to bridge the gap between IT and business services to enable IT services to run business services more effectively and efficiently. Web services have so far been the key technology for delivering on service computing.^4,15,16

Service computing aims to support the creation and delivery of services, which involve computing devices and software components distributed on the Web and provisioned (and often also controlled) by diverse organizations. Automated composition seeks to mash up these resources to provide customized IT services based on users’ goals and preferences.²⁴ To achieve this goal, standardization bodies, such as the World Wide Web Consortium (W3C) and the Organization for the Advancement of Structured Information Standards (OASIS), have led specification and standardization efforts for implementing service systems. In academia, service computing has attracted intensive attention. A number of journals and conferences were founded to assist research development in this area. The accompanying table lists some major venues and their predominant research topics.^c A state-of-the-art article on service computing can also be found in Yue et al.²⁴

Nevertheless, the state of the art in service computing remains largely in the realm of research and its potential for wide deployment has largely been unfulfilled. One key impediment has been the confusion created by myriad of often-competing Web service standards, attempting to standardize every single aspect of the service life cycle. Moreover, existing Web service standards and technologies do not provide adequate support for the computing needs of key emerging areas, which include mobile computing, cloud computing, big data, and social computing. These are the four key technologies of the Third Platform named by IDC that are currently influencing the landscape of global business.^d The two impediments (that is, confusion around Web service standards and lack of support for emerging computing grand challenges) have hampered wider and quicker adoption of service computing.

We call for revisiting the practice of superimposing the service paradigm with the technology of delivery (that is, Web services). Although Web services may remain relevant to developing various applications, we call for a focus on their underlying service needs. In particular, we call for the development of a new service paradigm that encompasses the four technologies identified by IDC, along with the introduction of new technologies, rather than emphasizing the expansion of existing Web service standards and technologies.

An aim of service computing is to harness the power and simplicity of the service paradigm with its functional and non-functional components to build modular software applications and provide a higher level of abstraction for selecting and composing services, thus elevating them to first class object status. A Service Oriented Architecture (SOA) is a technology-independent framework for defining, registering and invoking services.¹⁵ Service computing, however, is broader than SOA and includes the upper level of business process modeling, management, and analysis to the lower level of service data management and analytics. SOA has become a core concept of service computing and provides the fundamental technologies for realizing service computing. The emergence of cloud computing, as a new service delivery model, has inspired researchers and practitioners to explore the use of service computing in the cloud. A large body of research has been devoted to how service-computing concepts are leveraged to facilitate the development of cloud applications. As already mentioned, service computing can benefit from and contribute to new directions inspired by the emergence of mobile computing, cloud computing, big data, and social computing, as exemplified in Figure 2. This manifesto outlines the directions.

Since the emergence of the concept of service computing, a number of position and survey articles have been published.^7,9,16,24 The present effort differs from previous attempts in two main ways. First, the driving factor in this work is to decouple service computing from the technologies to implement service-oriented systems that take full advantage of the promises and expectations of service computing. Second, this work emphasizes the contribution and in of service computing on emerging trends in computing.

The structure of this manifesto starts with analyzing the obstacles to service-computing implementations, followed by examining the major challenges in emerging areas of service computing. We then present a research roadmap to address the identified challenges and draw conclusions and provide our assessments of the prospects for success.

Back to Top

Challenges in Service Computing Research

According to our survey on the articles published in the aforementioned journals and conferences (see the table), current service computing research focuses mostly on seven problem areas: architecture, specification languages, protocols, frameworks, life cycle, quality of service, and the establishment of trust and reputation across the boundaries of autonomous enterprises. The problems are especially apparent in the aforementioned emerging service domains.

An often-overlooked strategic challenge in service computing is analyzing why service computing has not reached its full potential in the real world, and what needs to be done to change that. A barrier that hinders service computing research from being transformed into an effective solution for industry challenges has been the concomitant lack of an easy methodology to transform complex data processing problems into routine services and an easy methodology to solve complex service interactions via data-centric architectures, such as Hadoop. A significant challenge is enabling the seamless cooperation of multiple organizations working on different platforms to satisfy consumers’ requirements.

---

What distinguishes services from other computing paradigms is their ability to work in a competitive environment where the key parameter to distinguish between similar services is their quality.

---

This would deliver on the much-needed service composition. Barriers to automated interorganizational service composition include disparities in competency, trust, accountability, functional and non-functional goals (including security and privacy) at the organizational level, and in interaction protocols and representations at the technical level.

We identify four emerging research challenges in service computing: Service design, service composition, crowdsourcing-based reputation, and the Internet of Things (IoT). First, we introduce service design, which is a fundamental but yet unsolved research problem in service computing. Second, we elaborate on challenges of service composition in the context of large-scale Web and cloud service systems, big data, and social networks. Next, crowdsourcing is a cost-effective means for IoT deployment, through collecting sensing data from pervasive sensing devices, for example, mobile phones.¹⁴ We focus on crowdsourcing as a key mechanism for reputation computation. Lastly, we discuss how service computing helps realize the IoT vision and challenges (see Figure 2).

Challenges in service design. Service design is about mapping a formal understanding of the nature of services and relationships thereof. This is an important prerequisite to building sound service systems. Service systems have so far been built without an adequate rigorous foundation that would enable reasoning about them. The preferred approach has been to rely on traditional software engineering approaches, which do not typically take into account the fact that service systems inherently bring together autonomous parts. The Web potentially provides a unique and uniform platform to develop sound service systems. However, there has not been any comprehensive theoretical framework to define and analyze complex service systems on the Web. As a result, only a few complex service systems have been developed.

Challenges in service composition. Partly fueled by the recent widespread successes of big data, the composition of large numbers of services into a coherent system is an emerging research area. We assess the scale involved as follows. In 2008, a survey discovered 5,077 WSDL-described Web services available on the Internet.¹ In 2016, there are almost 15,000 Web services registered in a public website.^e In addition, we believe that a large proportion of modern Web services are non-WSDL-described, such as those that are cloud-based. The current popularity of cloud computing has inspired the fast growth of cloud-based services and a 2013 survey discovered 6,686 cloud services.¹³ In the era of smart-phones, millions of apps are downloadable from cloud-based app stores. It is estimated that there are 1.6 million Android apps and 1.5 million Apple apps as of July 2015.^f Precisely and efficiently searching for services from these large-scale repositories is becoming a critical challenge.^2,24

In the setting of big data that might be accessed simultaneously by numerous services, existing service selection, composition, and recommendation approaches that mostly assume a static data environment are inadequate. Composition technologies that address consistency should be explored. In the setting of the IoT, Smart City can be viewed as a typical example of large-scale service composition, where millions of diverse and heterogeneous digital devices and services are integrated dynamically for provisioning multiple real-time functions or user-customized functions. Selecting and composing services from such numerous and ever-changing devices and services to fulfill user requirements in a real-time and context-aware fashion is a difficult mission, which requires urgent attention.²²

In large social networks, such as Facebook and Twitter, in which billions of users register and most users have hundreds of friends or followers on average (on average, 338 friends per user in Facebook^g in 2010 and 208 followers per user in Twitter^h in 2013), the resultant big data is complex as well as large. Service composition based on social relations poses fundamental serious challenges.

Challenges in crowdsourcing-based reputation. Trust plays an important role in the functioning of a service ecosystem. It is, however, difficult to establish trust when services that offer similar functionalities compete. Reputation is an effective approach in social networks for predicting credibility based on past behavior gleaned from consumers. It is often difficult to ascertain reputation in an open and often anonymous environment. Hence, reputation and crowdsourcing are important approaches for deriving trust.

Computing the reputation of a service entity in a community is realized by collecting all individual users’ opinions toward this entity in this community. Crowdsourcing provides an effective means for data collection through collaborations within the community.⁵ Nevertheless, several research challenges remain in the computation of crowdsourcing-based reputation.

Quality of crowdsourcing. The major difference between crowdsourcing and traditional user feedback collection is that crowdsourcing is more likely to use financial rewards and other incentives to motivate participation. However, it is unclear how these factors influence the quality of crowdsourcing. In addition, service users’ opinions are typically ambiguous, context-dependent, and based on individual preference. Ambiguity refers to what users express instead what they really think. Users’ opinions may also be affected by time, location, and social factors (for example, social relations between service users or between service users and service providers.¹² Some users may have particular preferences on certain service products.¹⁰ There is a strong need to predict the outcome of crowdsourcing reputation given that the reputation is influenced by several dependent factors.

Trustworthiness of crowdsourcing contributors. Some service users’ opinions might be unfair and even malicious towards particular service products. Unfairness or maliciousness is context-dependent, affected by the variation in services, time, and locations. A key issue is the selection of crowdsourcing contributors based on their trustworthiness, taking into account the context in which they operate.

Testing platform. Hitherto, there is no standardized testing platform to compare trust and reputation models. There is a strong demand for designing appropriate evaluation metrics to compare trust and reputation models for services.

Challenges in the IoT. The Internet of Things (IoT) is an emerging and promising area that proposes to turn every tangible entity into a node on the Internet. More specifically, the tangible entities (“things”) are any Internet-connected sensor, camera, display, smartphone, or other smart communicating devices. The IoT poses two fundamental challenges:¹⁷ communication with things, and management of things. Service technologies can help address these challenges, through provisioning communication and interoperability means to the IoT, such as REST and service composition models. One challenge is that things are resource-constrained and traditional standards, such as SOAP and BPEL, are too heavyweight to be applicable in the IoT. Furthermore, existing models of service composition cannot be directly used for IoT interoperation, because of architectural differences. In contrast to the single-type Web service component model (such as, services), the IoT component model is heterogeneous and multilayered (for example, devices, data, services, and organizations). Innovative models are required for IoT composition. Specifications of the functionalities of the IoT are a key challenge. Through the diverse, mobile, and context-aware devices, data and services can simultaneously generate diverse and context-aware functions.¹¹ Hence, in the IoT, desired functionality of components is more dynamic and context-aware than in traditional settings, which brings significant complexity in the composition process.

In a nutshell, the IoT involves incorporating possibly billions of things, and harnessing their data and functionality to provide novel smart services that benefit enterprises, industries, and our society. The fundamental IoT challenges as related to service computing are:

1. Continuously maintaining cyber personalities and contexts for IoT devices. In particular, IoT things need to have Web identities and Web representations (for example, Web proxies) that reflect their physical spaces. They also need to connect and communicate within social, environmental, user-centric, and application contexts, and such contexts need to be maintained and managed.
2. Continually discovering, integrating, and (re)using IoT things and their data. Specifically, the IoT environment is a federated environment where things and their data, cloud services, and IT services (for example, for data analysis and visualization) are often provided by independent providers with diverse interfaces, as well as business, cost, and QoS models. To provide new Internet-scale services, the IoT must (re)use IoT things deployed by others and data collected by others for their own purpose.

Back to Top

Service Computing Research Roadmap

Following the major challenges identified earlier, we now propose a research roadmap for service computing. The roadmap focuses on four emerging research areas, namely service design, service composition, crowdsourcing-based reputation, and Internet of Things.

Service design. Research should focus on laying the foundation for service design drawing from similar research in databases, software engineering, and distributed systems. The design of service systems should build upon a formal model of services that enables efficient access to a large service space with diverse functionalities. The service model can support accessing services much as the relational model supports accessing structured data. As service composition is usually needed to fulfill complex user requests, it is important for the service model to support dependency relationships among different services and their operations. Similar to the relational model, the service model should enable the design of service query algebra and calculus that allow general users to declaratively query multiple services in a transparent and efficient manner. In addition, service query optimization should go beyond just generating efficient query execution plans. As a large number of service providers may compete to offer similar functionalities, the design of the service model should support the implementation of optimization strategies for finding the “best” Web services or composition thereof with respect to the expected user-supplied quality. In sum, a formal service model that meets the requirements noted here will take center stage to provide efficient and transparent access to computing resources through services, which is a critical step to reach the full potential of service computing.

Recent work²⁵ has shown great promise in designing and using a formal service model to help users access service. A set of algebraic operators, defined based on a formal service model, provides standard ways to manipulate services as first-class objects. These include choosing specific operations from a service as well as quality-based service selection and mashup operators across multiple services to form a composed service. The implementation of these operators enables the generation of service execution plans that can be directly used by users to access services. Due to functional dependency and quality dimensions, a service model is more complex than a relational data model.

In particular, three key features of services are crucial: functionality, behavior, and quality. Functionality is specified by the operations offered by a service; behavior reflects how the service operations can be invoked, and is decided by the dependency constraints between service operations; quality determines the non-functional properties of a service. A viable service query language should allow users to locate and invoke their desired functionalities, select the best service providers that meet their quality requirements, and generate service compositions when functionalities from multiple services are required.

An important approach to design a composition of services is to do so not from the perspective of one party, but from a representation of interactions between two or more autonomous parties. Protocols are specifications of interactions that assiduously disregard the implementations of the services, but describe the interactions: protocols thus promote interoperability among heterogeneous and autonomous services.²³ Newer approaches for protocols accommodate data representations, but additional research is needed to bridge the gap between interaction and data. An important direction is to formalize sociotechnical interactions as a basis for governance of autonomous services.²⁰

Service composition. Future research in service composition should target the following topics:

Large-scale Web and cloud service composition. Service composition research should extend to non-WSDL-described services or services described by plain text. As an example, ProgrammableWeb.com hosts over 10,000 API services and over 6,000 service mashups, with a great majority described in plain text. Web information extraction, natural language processing, data and text mining, collaborative tagging, and information retrieval technologies can be used to extract useful semantics, group relevant services, and detect novel composition patterns. Nevertheless, short service descriptions comprised of limited terms coupled with the diverse naming conventions used by service providers pose novel challenges that demand technological innovations to advance the state of the art in both service computing and all relevant domains. Furthermore, a software system may need to evolve during its life cycle and be able to handle the changes in its running environment and the increasing complexity of its workflows. Self-adaptive software evaluates its behavior and makes adjustment according to the evaluation result to address issues and improve performance.

A cloud-computing environment provides an attractive option for deploying services, because of the potential scalability and accessibility it offers. However, it introduces problems related to:

1. Maintenance—The resources are not under the explicit control of the service provider.
2. Security—The cloud might not be within the enterprise boundaries of the service provider.
3. Service-Level Agreements (SLAs)—Resource allocation is the responsibility of the cloud provider. For example, a service might be unavailable not only due to updates by the service provider, but also due to updates by the cloud provider.

These problems must be addressed as services are moved to clouds and, recently, containers within clouds.

Big data driven service composition. Due to limitations of existing methods, new service selection, composition, and recommendation technologies are key approaches to leveraging state-of-the-art big data research outcomes. One important topic in current big data research is the development of algorithms and models for processing data online. This is fundamentally different from the traditional batch processing approach. Online service composition may provide a promising direction to achieve scalable and adaptive composition solutions to deal with large-scale, highly dynamic, and diverse big data services. Another important consideration is how humans interpret the outcomes of service computing and big data frameworks. As such, it will become increasingly important to relate big data analytic frameworks with the nature of the humans for whom they serve.²¹

Social-network-based service composition. Service selection, recommendation, and composition in large-scale social networks should target combining social network and complex network analysis methods, as well as trust computing techniques. One promising direction is to incorporate social network data that records the interaction of service users with the service data to detect hidden relationship between services and generate potential service compositions. In particular, user activities exhibited through social media services, such as posting experiences, questions and feedbacks about services, could bring novel insight to better understand and leverage both traditional and emerging services. For example, the records of using services to build service mashups posted on public programming forums can be leveraged to recommend interesting services to other users and build new service mashups. As another example, user behavior patterns can be detected from event logs recorded by social media or e-business platforms and used to discover latent knowledge for business process mining. Furthermore, emerging services may bring novel QoS features that go beyond the traditional ones, such as reliability, availability, and response time. Domain-specific quality features that reflect users’ interests in choosing and composing services could be extracted through social media services that capture user personal judgment.¹⁰

---

The design of service systems should build upon a formal model of services that enables efficient access to a large service space with diverse functionalities.

---

Crowdsourcing-based reputation. Future research on crowdsourcing-based reputation of services should target the following directions:

Quality of crowdsourcing. Future studies should focus on how monetary or other interesting factors affect the quality of crowdsourcing data for choosing services. Social studies should be carried out to survey the impact of these interest factors on the reliability of crowdsourcing and the scope of the crowdsourcing contributors. The three factors (ambiguity, context-dependency, and individual preference) that influence the quality of crowdsourcing data should be studied. For the ambiguity issue, future research should focus on how trust assessment questions and scales as well as human computer interactions should be designed to precisely capture users’ trust-related perceptions, through the collaboration with cognitive science and human-computer interaction research. The difference between context-dependency and individual preference is that the former relies more on the objective factors, for example, time, location, and social relations, and the latter relies more on the subjective factors, for example, personal experience. These two kinds of factors, for example, social relations and personal preference, may influence each other simultaneously. Future research should target how to model the interrelation between the two groups of factors and how they can be integrated to predict their impact on the quality of crowdsourcing data.

Trustworthiness of crowdsourcing contributors. Future research should focus on studying the context-dependent features of contributors’ trustworthiness. The credibility of a crowd member (that is, a human user or another service) determines how much other service consumers may trust its reported ratings regarding the services it has invoked. This allows us to differentiate between service trust and feedback trust. For instance, a service that is not trustworthy as a service provider may be trustworthy as a judge of the behavior of other service providers, and vice versa. Tradeoff strategies for selecting users with different cost and trustworthy levels for crowdsourcing should also be explored. Selecting the most appropriate crowdsourcing contributors or crowd workers will require services that interact and combine information from contributors, service providers, and third-party sources, such as job listings and social media.⁸

Internet of Things. In traditional service computing, a service composition focuses on finding an effective combination of component services.¹⁹ Recent work has suggested that a service composition in the IoT needs to find an effective combination of both component services and devices.¹¹ In emerging IoT architectures based on service composition, there is a need to find an effective combination of component services, data services supported by cloud platform services, and devices. Cloud component services are an integral part of the IoT, because they are needed to manage device Web representations, contexts, and related data processing services anywhere and without the need to rely on a computing center.

Complementary to the current work on service discovery and integration, an important and novel direction in IoT research lies in the area of device discovery and integration. Existing results based on a combination of Semantic Sensor Networks (SSN) and OpenIoT¹⁸ provide a device layer architecture and related functionality for IoT device discovery and integration. SSN defines an ontology that is used to describe IoT things and to find devices from the attributes of the data they produce. Nevertheless, as IoT things are diverse in functionality and access methods, it is infeasible for a single uniform ontology to cover the highly heterogeneous space of things. Information retrieval and text mining techniques can be leveraged to improve the accuracy of device discovery.

Integration of IoT things can be greatly facilitated by discovering the correlations among things. However, correlations among IoT things are usually difficult to detect. Unlike human social networks, where people are well connected, things often have limited explicit interconnections. An interesting direction is multi-hop connections, which leverage human-to-things interactions to correlate IoT things.

Graph-based approaches and machine-learning techniques can help discover hidden interesting relationships among IoT things, and thus suggest interesting and novel integration patterns.

Back to Top

Conclusion

Service computing has a bright future supporting the tremendous advances in emerging areas of computing such as mobile computing, cloud computing, big data, social computing, and beyond. We make the case in this manifesto that the potential of service computing is far greater than what has been achieved so far. We lay down a path forward to take service computing to new heights of innovation. To forge ahead, we make the important statement that, for the service computing paradigm to succeed, it needs to be decoupled from the technology of the day. The challenges may be difficult but the payoffs are great and there is no reason why an ambitious research agenda would not produce enormous benefits for computer science and society.

Back to Top

Acknowledgments

This research was made possible by LP120200305, DP150100149, and DP160103595 grants from Australian Research Council and NPRP 7-481-1088 and NPRP 9-224-1-049 grants from the Qatar National Research Fund (a member of The Qatar Foundation). The statements made herein are solely the responsibility of the authors.

Back to Top

Back to Top

Back to Top

Back to Top

Back to top