Cloud services (CCS) are a crucial element in the service sector, but there are still challenges left in their design, implementation, operation and dismissal, due to issues such as the integration of physical and technical components, interaction of social and technical aspects, dynamic and elastic recon guration. In modelling service systems, ontologies have been recognized as a useful instrument for reducing conceptual ambiguities and inconsistencies. However, none of the general approaches proposed in literature have addressed the speci c aspects of CCS. In this perspective, we explore how the UFO-S core ontology can be used to describe IT services and, in particular, CCS. A case study and the challenges deriving by CCS are discussed.
For many years, services have been investigated from the economic, nancial and juridical viewpoints due to their socio-economic relevance. More recently, IT services have become a very active eld of study, due to the complex interplay among the above-mentioned aspects and the peculiar features of the digital milieu (e.g. ubiquity, mobility, context-awareness). Cloud computing services (CCSs) di er from traditional IT services for some characteristics, such as abstraction from the underlying hardware/software infrastructure, multi-platform accessibility, on-demand service provisioning, pay-per-use-based business models, dynamic quality of service (QoS) management, scalability and exibility. These peculiarities, which are responsible for the rapid emergence of a large number of CCSs, motivate the importance of a speci c analysis.
From the customer perspective, \the great amount of CCSs makes it hard to
compare the o ers and to nd the right service" [11, p. 81]. The customer needs
to evaluate service features and expected outcomes, their correspondence to his
needs, the risks associated to the service (e.g., lock in, security) and the
opportunities (e.g., integration with other services). Starting from these considerations,
the customer can choose which service to buy. This information is also useful
to the provider for strategic purposes; for instance in order to evaluate possible
service compositions, service pricing, or the dynamic allocation of resources [
It seems clear that a proper analysis of CCS from the conceptual perspective requires a holistic approach, taking the whole service system into account. This means understanding target customers, relations among actors, and the speci c ways of value co-creation. In turn, this demands understanding the way in which actors operate, interact and use resources to co-create value.
In modelling service systems, ontologies have been recognized as a useful
instrument for reducing conceptual ambiguities and inconsistencies [5, 17, ?, 16].
None of these general approaches, however, have addressed the peculiarities of
CCS. In this paper we would like to explore how a general ontology of services {
the core ontology of services developed in [
Our aim is to identify which are the changes needed to UFO-S in order to account for CCS and which are the aspects that can be already modelled using the existing primitives. This will be based on the application of UFO-S to a speci c case study, which { besides the immediate relevance to the cloud computing domain { represents a particularly complex domain.
The remainder of the paper is structured as follows. In Sec. 2, we outline the peculiarities of CCS, thus setting the groundwork on why these services pose more challenges than other IT services. In Sec. 3 we shortly recap UFO-S, which we use to model a case study (Sec. 4). Finally, based on the previous sections, we outline the main modeling challenges for IT services (Sec. 5).
Cloud services are based on the cloud computing technology, which has been
de ned by the National Institute of Standards and Technology (NIST) as \a
model for enabling ubiquitous, convenient, on-demand network access to a shared
pool of con gurable computing resources [. . . ] " ([
The heterogeneity in CCSs is so signi cant [
Service deployment is a complex process that undergoes several phases
(collectively known as service lifecycle), such as service design, service
implementation, service o ering, service negotiation and agreement, service delivery, service
support and service end-of-life management [
On a rst approximation, there are four main roles involved, namely service
provider, service producer, service customer and service consumer [
Providers can be either application providers, platform providers or
infrastructure providers. The application provider provides applications to customers
and is responsible for overall service monitoring and quality assurance. The
platform provider o ers \an environment to develop, run and test applications " ([
Intermediate roles are played by consultants, aggregators and integrators. Consultants o er expertise on cloud computing and on the customer's business processes and requirements. Aggregators assemble CCS in order to provide more complex solutions. Integrators intervene when there is not a previous aggregation of CCS, but it is the customer that decide which services to integrate. Finally, consumers are the direct bene ciaries of the service.
Cloud services are characterised by high dynamicity from di erent points of
view. We can distinguish the requirement layer (from the customer/consumer
perspective), the resource layer, the value layer and the legal layer. From the
customer perspective, requirements towards service functionalities { based on
customers' goals, which can rapidly change { are the basis for choosing among
services and di ers based on the user community, which may have di erent needs.
From the provider perspective, the dinamicity in the demand of the service brings
to the dinamicity at the resource layer, in terms of capabilities required to
design, develop and deploy the service. In accordance with with the resource-based
view theory [
Based on CCS characteristics and their dynamism, it is necessary to account for: (a) the di erent roles that actors can assume (e.g., an organization is a customer of a company and provider of another one); (b) the characteristics of the actors involved, the external environment (e.g., competitors) and the internal structure and dynamics of the organizations, also in terms of resources used.
UFO-S, a core reference ontology for services, is able to explain a number of
perspectives on services, including those that emphasize services as value
cocreation, as capabilities and as application of competences [
As a core ontology [
First of all, each object is considered as an instance of a kind, which is a
substantial sortal universal. Each entity can assume a role depending on the context.
An entity capable of covering many other concepts with di erent principles of
identity is considered as a mixin. Other kinds of types can be highlighted: phases
represent possible stages in the history of a substance sortal (e.g., for a living
thing, alive and deceased). In the same way, modes are individuals existentially
dependent on other individuals.Other basic concepts include: agents (e.g.,
person, organization); physical or social objects, i.e., non-agentive substantial
particulars; actions, which stand for intentional events whose existence depends on
their own participants; resources, i.e., objects participating in an action. In
addition, a crucial role in UFO-S is played by relators, which can be seen as rei ed
relationship. More exactly, relators, whose ontological nature and signi cance
for the practice of conceptual modelling has been recently revisited in [
The UFO stereotypes sketched so far allow us to understand how service
lifecycle phases have been modelled in [
In this paper, we focus on two phases of the service life-cycle according to the
formalization proposed in [
Let us see now how a reference ontology of services, UFO-S, can be applied to develop a service ontology concerning a concrete example of a cloud service, coming from a real case study. The case study pertains an email service internally delivered to an Italian company with more than 5000 employees spread out into more than 100 o ces all over the country. The IT Department of such company was responsible for procuring this service. After a public call, two service providers were selected for the mailbox service and the networking service.
The model we developed resulting from the application of UFO-S to our case study is reported in Fig. 1. To better understand the scenario, we have divided our model in two parts, representing what happens at the contractual level (upper yellow layer) and at the delivery level (bottom green layer). The central entities in these two parts are respectively the IT Service Contractual Relationship and the IT Service Factual Relationship. This structure re ects a peculiarity of our domain, and in particular of CCS: during the service delivery phase the actual resources allocated by the provider are dynamically adjusted, and contractual aspects keep being dynamically re-negotiated while the service delivery evolve. So we have two relationships that evolve more or less at the same time: a contractual relationship and a factual relationship. Thanks to the relator construct we can account for both. In particular, as we have shown in the model, we can account for di erent phases of the contractual relationship, such as, for example, a test phase where the optimal resources to be allocated are estimated, a normal phase, and an emergency phase where for some reasons there is a scarcity of resources, and some priority policies need to be adopted on the basis of the customer's needs. Concerning the customer's needs, we can see that they are represented as a mode inhering in the customer, but which depends on a speci c department inside the customer's organization, namely the IT department. Together with the customer's and the providers' commitments, customer's needs are part of the bundle that constitute the factual relationships, whilst only the rst two are relevant for the contractual relationship.
Both relationships involve the Hired ICT Provider, specialized in the network and mailbox service provider, and the Business Customer. It is important to highlight that although it is the business customer who is bound to the contractual and factual relationships, it is the IT Department who participates in the Initial Service Negotiation. Thus, it establishes commitments and claims on behalf of the business customer. As a consequence, the commitments and claims established by the IT Department "belongs" to the business customer.
It is worth noticing that the core action in the service delivery, namely the Single Mail Action, is not performed by the providers but rather from the User (see Sec. 5). The user is, at the same time, the bene ciary of the service, though is not the one that can choose which providers to hire or under which conditions the service is delivered (i.e., the IT department), nor the one who actually pays for the service (business customer). Thus, the user is the consumer of the service. The Hired Network Service Provider performs the action of providing the Internet connection, by allocating the required Internet Bandwidth.
We must observe that, despite we have tried to use UFO-S as much as possible, the peculiarity of our case study has forced us to deviate from it in many respects. One aspect concerns the relation between the provider's commitment and the action that constitutes what the provider commits on, that is, in our case, a mail sending/receiving action, or a mailbox management action. This action is guaranteed by the provider, but it is actually executed by the user, which is in this case a customer's employee. In this particular case, the action on which the mailbox provider commits presupposes another action, namely some kind of internet transport (or internet connection), which is guaranteed by another provider: the network provider.
As we have seen, conceptual modelling of CCS poses several challenges, parts of which can be addressed with the current version of UFO-S and parts of which require its extension. Let us recap them.
The analysis of the CCS characteristics and of the case study brings to light how the corresponding models need to re ect dynamism and exibility of CCS, in terms of both service structure and governance choices based on a cost-bene t analysis. The relational dynamic aspects can be modeled by means of relators since they are bundles of qualities { in this case, commitments and claims { that account for the way in which the related entities are involved in the relationship.
Indeed, it is necessary to distinguish among what is de ned contractually, i.e. the contractual relation, and what is actually done, i.e., the factual relation.
As previously stated, in the service delivery phase it may be necessary to dynamically re-negotiate contractual conditions and allocate resources. The former aspect concerns the rede nition of the contract based on the actual commitment of the providers and on the customer needs expressed by the IT department. The contractual relation is accounted for in UFO-S, while the factual relation is not yet factored in. The latter trait, together with the dynamic optimization of resources, is one of the key aspects of cloud services. This peculiarity requires to account for both the resources and the value of the service, in order to face the trade-o between costs, QoS, potential revenues and penalties. A change in the allocation of the resources, may bring to the payment of penalties or to revenue loss. The potential revenues can be analyzed by understanding how value is perceived by the target customers, thus more e ectively tailoring the price to the context in which the service is sold and on the characteristics of the target customers. In this sense, it should be possible to model organizations, societal aspects, target customers, high-level preferences. Among the aspects that may a ect customers' preferences , there are the lock-in risk, software license type, privacy and security concerns, which can be modeled with resources, besides the more common considerations of the actions needed to buy or use the service.
The dynamic allocation of resources a ects also the relevance of the notion of
commitment, on which UFO-S is built. Indeed, the customer pays for having the
provider committed to procure the amount of resources needed by the user in
order to bene t from the service [
To consider also the aspects related to value propositions and contractual issues, the di erent phases of the service lifecycle have to be analyzed and extended, from the service design to the o ering, the delivery and, nally, the termination phase. In the latter phase, data storage issues gain particular importance for CCS, since the provider is in charge of deleting all customer-related data after service termination. The concept of commitment is useful to model this traits, as well as all other obligations assumed by the provider towards the customer. These aspects a ect value as well, as customers evaluate potential data breaches, penalties in case of cancellation of the contract and several other aspects to decide how valuable the service can be for itself or for its company.
Moreover, it is important to tackle role changes for a given actor, especially in service chain scenarios. If we refer to the proposed case study, we have that a given company buys two services from two di erent external providers in order to integrate them and o er them to its employees. Thus, the company o ers the integrated IT service to its employees, without being responsible for the quality of these services (QoS) or being able to intervene on the delivered QoS. In this sense, the company is not only a customer but rather a service aggregator that, from the employees' point of view, behaves as a service provider, thus denoting the modelling need of role changes for actors.In addition, another issue related to roles has to be mentioned as well. As highlighted in Sections ?? and 4, besides the roles of service provider and customer, other ones should be accounted for, such as service consumer and service aggregator.
Finally, IT services { and in particular CCS { are seldom instrumental
services. With instrumental service we mean that the o er of the provider does not
consist of an action (e.g., cutting your hair), but rather of allowing the user to
perform a given action, which constitute the core action of the service. In our
case study, the providers o er the Internet connection and the mailbox
application, with whom they guarantee to the users that they can send/receive or
manage emails. In this frame, the provider performs supporting actions apt at
enabling the core service consumption [
This work investigate whether cloud services can be represented by means of the core ontology for services, namely Uni ed Foundational Ontology for Services (UFO-S). In order to do so, we outline CCS peculiarities, e.g., dinamicity, and the roles of the actors involved in their deployment and usage. Thus, we apply UFO-S to a case study concerning the external provisioning and internal delivery of an email service in a big company.
Through the modeling of the case study, we outline the main bene ts of UFO-S and the extension required to model CCS, focusing on the necessity of representing dynamism, value, roles and actions . Besides the previously described general advantages of the adoption of UFO-S, we analyze the relavance of relators for CCS modeling. Relators are a bundle of qualities, through which it is possible to represent dynamic relationships among providers and customers in the lifecycle phases e.g. contractual and factual relationships. In the current version of UFO-S only the initial agreement relationships is factored in, while there is no account, e.g. for the factual relationship. Thus, an extension is needed.
We also show that in order to account for value, value propositions and contractual aspects the phases of the service lifecycle need to be expanded, so to include the service design and termination phases. The complexity of CCS requires for an in-depth analysis of roles, which can be multiple for the same actor. At the moment, it is not possible to represent this with UFO-S. The complexity of CSS requires for an in-depth analysis of roles. Thus, the number of service participant roles should be extended in UFO-S in order to include also the ones de ned in cloud computing literature. Finally, we highlight how the analysis of the notions of core and supporting actions is necessary in order caractherize instrumental services.
Future work will be directed at the integration of these aspects in UFO-S, with the aim of providing a new version of the core ontology.