Service-Oriented Distributed Applications in the Future Internet: The Case for Interaction Paradigm Interoperability

TLDR: This paper introduces an interoperability solution based on abstraction and merging of the common high-level semantics of interaction paradigms, which is sufficiently general and extensible to accommodate many different protocol technologies.

Abstract: The essential issue of interoperability in distributed systems is becoming even more pressing in the Future Internet, where complex applications will be composed from extremely heterogeneous systems. Open system integration paradigms, such as service oriented architecture (SOA) and enterprise service bus (ESB), have provided answers to the interoperability requirement. However, when it comes to integrating systems featuring heterogeneous interaction paradigms, such as client-service, publish-subscribe and tuple space, existing solutions are typically ad hoc and partial, applying to specific interaction protocol technologies. In this paper, we introduce an interoperability solution based on abstraction and merging of the common high-level semantics of interaction paradigms, which is sufficiently general and extensible to accommodate many different protocol technologies. We apply this solution to revisit the SOA- and ESB-based integration of heterogeneous distributed systems.

Figures

Fig. 7. Binding Component architecture.

Fig. 2. CS IDL

Fig. 4. TS IDL

Fig. 3. PS IDL

Fig. 6. eXtensible Service Bus.

Table 3. Interaction latency on the bus for each interconnection

Full text

HAL Id: hal-00841332
https://hal.inria.fr/hal-00841332
Submitted on 4 Jul 2013
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Service-oriented Distributed Applications in the Future
Internet: The Case for Interaction Paradigm
Interoperability
Nikolaos Georgantas, Georgios Bouloukakis, Sandrine Beauche, Valérie Issarny
To cite this version:
Nikolaos Georgantas, Georgios Bouloukakis, Sandrine Beauche, Valérie Issarny. Service-oriented Dis-
tributed Applications in the Future Internet: The Case for Interaction Paradigm Interoperability.
ESOCC 2013 - European Conference on Service-Oriented and Cloud Computing, Sep 2013, Malaga,
Spain. pp.134-148, �10.1007/978-3-642-40651-5_11�. �hal-00841332�

Service-oriented Distributed Applications in the
Future Internet: The Case for Interaction
Paradigm Interoperability
Nikolaos Georgantas
1
, Georgios Bouloukakis
1
, Sandrine Beauche
2
, Val´erie
Issarny
1
1
Inria Paris-Rocquencourt, France
firstname.lastname@inria.fr
2
Ambientic, France
sandrine.beauche@ambientic.com
Abstract. The essential issue of interoperability in distributed systems
is becoming even more pressing in the Future Internet, where complex
applications will be composed from extremely heterogeneous systems.
Open system integration paradigms, such as service oriented architec-
ture (SOA) and enterprise service bus (ESB), have provided answers to
the interoperability requirement. However, when it comes to integrating
systems featuring heterogeneous interaction paradigms, such as client-
service, publish-subscribe and tuple space, existing solutions are typically
ad hoc and partial, applying to specific interaction protocol technologies.
In this paper, we introduce an interoperability solution based on ab-
straction and merging of the common high-level semantics of interaction
paradigms, which is sufficiently general and extensible to accommodate
many different protocol technologies. We apply this solution to revisit the
SOA- and ESB-based integration of heterogeneous distributed systems.
Key words: Interoperability, interaction paradigms, interaction abstrac-
tions, service oriented architecture, enterprise service bus.
1 Introduction
The Future Internet (FI) is emerging as, among others, a global application
space where People, Services and Things will be always-connected and interact
in numerous ways. Accordingly, complex distributed applications in the FI will
be based, to a large extent, on the open integration of extremely heterogeneous
systems, such as lightweight embedded systems (e.g., sensors, actuators and net-
works of them), mobile systems (e.g., smartphone applications), and resource-
rich IT systems (e.g., systems hosted on enterprise servers and Cloud infrastruc-
tures). These heterogeneous system domains differ significantly in terms of in-
teraction paradigms, communication protocols, and data representation models,
which are most often provided by supporting middleware platforms. In partic-
ular with regard to middleware-supported interaction, the client-service (CS),
publish-subscribe (PS), and tuple space (TS) paradigms are among the most

2
widely employed ones, with numerous related middleware platforms, such as:
Web Services, Java RMI for CS; JMS, SIENA for PS [1, 2]; and JavaSpaces, Lime
for TS [3, 4]. In the following, we outline a representative application scenario,
where a complex distributed application needs to be devised by integrating het-
erogeneous networked systems that interact with differing interaction paradigms.
Search and Rescue (S&R) operations after a disaster, such as a flood or earth-
quake, are carried out in hazardous environments and require personnel from
multiple agencies (e.g., fire-fighters, police) to coordinate. To detect survivors,
sensors are installed at various places of the hazardous area. Such sensors com-
municate their location. S&R personnel also notify at short intervals of their
current positions via their PDAs. Upon sensing some life sign, sensor nodes
send out notifications. At the same time, nearby light-emitting actuators start
lighting the place to facilitate the rescuing effort. Sensors, PDAs, and actuators
interact among them and with external actors via a TS. TS location and life sign
data are sent via CS invocations to a planning service that recommends at real
time the optimal deployment of rescue forces. This output is notified via a PS
system to the coordinator of the operation on her smartphone and also to a num-
ber of control/monitoring centers. The coordinator may approve and command
S&R personnel via the PS system and the TS system to rush into the spot.
To enable such a scenario, the heterogeneity between the involved system do-
mains needs to be tackled. Existing cross-domain interoperability efforts are
based on, e.g., bridging communication protocols [5], wrapping systems behind
standard technology interfaces [6], and providing common API abstractions [7–
10]. In particular, such techniques have been applied by the two currently domi-
nant system integration paradigms, that is, service oriented architecture (SOA)
and enterprise service bus (ESB) [11]. Both SOA and ESB employ the CS
paradigm. Certainly, there are extensions, such as event-driven SOA [11] or
industrial-strength ESBs supporting the PS paradigm. Additionally, research
efforts have proposed the TS paradigm as interaction substrate for Web services
or for ESBs [9, 12]. Nevertheless, most of these cross-paradigm interoperability
efforts are ad hoc and partial, applying to specific cases. On the other hand,
interaction paradigms have been widely studied, with theoretical approaches
providing them with formal semantics by relying on concurrency theory, process
algebras and architectural connectors (e.g., see [13]). These approaches typically
identify semantics for individual paradigms but not cross-paradigm semantics.
In this paper
3
, we introduce a model-based system integration solution that
can deal with diverse existing systems, focusing in particular on integrating their
heterogeneous interaction paradigms. Our systematic approach is carried out in
two stages. First, a middleware platform is abstracted under a corresponding
interaction paradigm among the three base ones, i.e., CS, PS and TS. To this
aim, we elicit a connector model for each paradigm, which comprehensively cov-
ers its essential semantics. Then, these three models are abstracted further into
a single generic application (GA) connector model, which encompasses their
3
This work has been partially supported by the European Union’s Seventh Framework Programme
FP7/2007-2013 under grant agreement number 257178 (project CHOReOS).

3
GA connector
app A
app B
PS
connector
TS
connector
connector
converter
CS
connector
app C
Fig. 1. GA-based connector interoperability
common interaction semantics. Based on GA, we build abstract connector con-
verters that enable interconnecting the base interaction paradigms. A high-level
representation of our approach is depicted in Figure 1. We realize our interoper-
ability solution as an extensible service bus (XSB), which is an abstract service
bus that employs GA as its common bus protocol. Furthermore, we provide an
implementation of the XSB, building upon existing SOA and ESB realizations.
Based on our XSB platform, we propose a comprehensive solution to the peer-
to-peer integration of services relying on heterogeneous interaction paradigms
into complex applications. Our overall approach generalizes the way to design
and implement service-oriented distributed applications, where the employed
interaction paradigms are explicitly represented and systematically integrated.
We demonstrate the applicability of our solution by implementing the scenario
introduced above, and evaluate it in terms of extensibility and performance.
The rest of this paper is structured as follows. In Section 2, we introduce our
connector models for abstracting and interconnecting interaction paradigms. In
Section 3, we present the application of our models to the XSB solution, as well as
its implementation. Then, in Section 4, we discuss the results of our evaluation.
We finally complement this paper with a comparison of our approach with related
work in Section 5, and conclude, also discussing future work, in Section 6.
2 Abstractions for Interaction Paradigm Interoperability
In this section, we identify the semantics of the three principal interaction
paradigms, i.e., CS, PS and TS, and elicit a connector model for each paradigm
(Section 2.1). Our modeling proposition is the outcome of an extensive survey
of these paradigms as well as related middleware platforms in the literature.
In a second step, we introduce our GA connector model, which enables cross-
paradigm interoperability (Section 2.2). Before getting into the specifics of each
connector, we briefly introduce in the following our global approach to connector
modeling and point out the specific focus of this paper.
Our models represent the essential semantics of interaction paradigms, con-
cerning space coupling, time coupling [14] and concurrency. Space coupling de-
termines how peer applications interconnected via the connector identify each
other and, consequently, how interaction elements (e.g., messages for a CS con-
nector) are routed from one peer to the other. Time coupling determines if peers
need to be present and available at the same time for an interaction or if the

4
interaction can take place in phases occurring at different times. Concurrency
characterizes the exclusive or shared access semantics of the virtual channel es-
tablished between interacting peers. These three categories of semantics are of
primary importance, because these are end-to-end semantics: when interconnect-
ing different connectors, we seek to map and preserve these semantics.
We represent interaction paradigm semantics in the connector’s abstract
API (Application Programming Interface). This API presents the programming
model supported by the connector and offered to the peer applications that use
the connector for their interaction. The API is a set of primitives expressed as
operations or functions supported by the middleware. This abstract API can be
refined to a specific middleware platform by mapping to the primitives and in-
corporating the data structures and types of the middleware platform. Besides a
connector’s API, we introduce an abstract interface description language (IDL)
for specifying the open interfaces of systems that rely on middleware represented
by the specific connector. Our IDLs are largely inspired from WSDL. We specify
the IDLs conceptually, while we have also implemented each one of them as an
XML schema document. Based on the flexibility of XML schema, an IDL can
be easily refined in order to enable the description of a concrete system that is
based on the connector, e.g., we can refine the abstract XML elements into the
precise data structures and types of the specific middleware and system.
Based on the informal identification of semantics as discussed in the previous,
we further specify the connector’s formal behavioral semantics in terms of LTS
(Labeled Transition Systems). This formal behavior specification focuses on time
coupling and concurrency semantics, while space coupling semantics is mainly
represented by the connector’s API and IDL. Additionally, we formally verify the
correctness of these behavioral specifications with respect to time coupling and
concurrency properties expressed in LTL temporal logic. This allows stating the
correctness of our base connector models with respect to the semantics that they
must have. This further enables identifying the semantics of the GA connector
derived from the interconnection of base connectors.
The focus of this paper is the application of our connector modeling and
analysis approach to the practical integration of heterogeneous services. Hence,
and due to space limitations, we introduce in the following sections our connec-
tors only informally – concentrating on their space coupling, time coupling and
concurrency semantics – and mainly in terms of their respective IDLs, which are
used to describe open interfaces of services.
2.1 Connector Models for Base Interaction Paradigms
This section introduces connector models for the CS, PS and TS paradigms.
Client-service connector. The CS connector model integrates a wide range of
semantics, covering both the non queue-based messaging and remote procedure
call paradigms. In terms of space coupling between two interacting peers, CS
requires that the sender must hold a reference of the receiver. With respect to
time coupling, both entities must be connected at the time of the interaction.

Frequently asked questions

Q1. What have the authors contributed in "Service-oriented distributed applications in the future internet: the case for interaction paradigm interoperability" ?

In this paper, the authors introduce an interoperability solution based on abstraction and merging of the common high-level semantics of interaction paradigms, which is sufficiently general and extensible to accommodate many different protocol technologies. 

Q2. What future works have the authors mentioned in the paper "Service-oriented distributed applications in the future internet: the case for interaction paradigm interoperability" ?

The authors apply their modeling abstractions to extend an SOA & ESB infrastructure for supporting development of complex applications by seamless peer integration of heterogeneous services. In their current and future work, besides publishing on the formal foundation of their interoperability approach, the authors aim to enrich their modeling abstractions with support for continuous interactions in addition to discrete ones. Continuous interactions are commonly found in data streaming protocols, which are increasingly important in the Future Internet, due to the vast spread of media content and sensor-generated data streams. Additionally, support for new middleware platforms, new ESB infrastructures, or even new interaction paradigms can be incorporated in a facilitated way thanks to their architectural framework. 

Q3. What is the focus of this paper?

The focus of this paper is the application of their connector modeling and analysis approach to the practical integration of heterogeneous services. 

Q4. What are the main components of the cross-domain interoperability efforts?

Existing cross-domain interoperability efforts are based on, e.g., bridging communication protocols [5], wrapping systems behind standard technology interfaces [6], and providing common API abstractions [7– 10]. 

Q5. What are the main reasons for the use of continuous interactions in the future internet?

Continuous interactions are commonly found in data streaming protocols, which are increasingly important in the Future Internet, due to the vast spread of media content and sensor-generated data streams.9 

Q6. What is the key interaction element in PS-IDL?

The essential interaction element in PS-IDL is event ; its interaction semantics denotes whether this event is published or received by the system in question and its lifetime, determined by lease. 

Q7. What are the common middleware platforms used in the future internet?

In particular with regard to middleware-supported interaction, the client-service (CS), publish-subscribe (PS), and tuple space (TS) paradigms are among the most2 widely employed ones, with numerous related middleware platforms, such as: Web Services, Java RMI for CS; JMS, SIENA for PS [1, 2]; and JavaSpaces, Lime for TS [3, 4]. 

Q8. What is the future of the internet?

The Future Internet (FI) is emerging as, among others, a global application space where People, Services and Things will be always-connected and interact in numerous ways. 

Q9. How much effort is required for the integration of a new middleware platform?

The authors see that considerably small effort, no more than 6% of the total effort, is required for the integration of a new middleware platform. 

Q10. Why do the authors introduce their connector models informally?

and due to space limitations, the authors introduce in the following sections their connectors only informally – concentrating on their space coupling, time coupling and concurrency semantics – and mainly in terms of their respective IDLs, which are used to describe open interfaces of services. 

Q11. What is the way to build composite applications?

Using this platform, they can easily develop composite applications: they only need to build descriptions for the constituent services and directives for data mapping among them. 

Q12. How many times do the authors compare the XSB latency overhead with the EasyESB?

When conversion between heterogeneous interaction paradigms is involved, the XSB latency overhead ranges from 7% to 15,5%, where the authors note that the authors always compare with the EasyESB CS-CS homogeneous interconnection, since EasyESB support for other interaction paradigms is not available. 

Q13. What is the comparative advantage of their approach?

With respect to these efforts, the comparative advantage of their approach is its generality and extensibility thanks to the introduction of the higher-level GA abstraction. 

Q14. What is the interaction semantics of a tuple?

Its interaction semantics denotes whether this tuple is produced or consumed by the system in question and its lifetime, determined by lease. 

Q15. What is the main new concept in the PS-IDL?

to introduce their PS-IDL (Fig. 3), the authors rely on their PS connector semantics, which has been extracted from a wide range of PS systems.