Conference PaperPDF Available

Components for Growing the RESTful Enterprise.


Abstract and Figures

For a modernenterprise, it is vital to be on the Web. Beyond offeringhuman-readable Web sites, organizations increasingly use the Web as a media formachine-readable data about itself. With the help of technologies like XML feeds,RESTful Web services and semanticmarkup, new forms of enterprisesmodelsemerge in a bottom-up way. These models are easily consumable and facilitate theinteraction with departments,partners and customers.Engineering good publishingsystems is howeverextremelychallenging.On the one hand, knowledge of manytechnologies is required; on the other hand, it must be easy to extendsystems anddata models in accordance to agile businesses. In this paper, we proposea frame-work of components for publishing dynamically growingenterprisemodelson theWeb, present an implemented system and discuss its use in a case study.
Content may be subject to copyright.
Components for Growing the RESTful Enterprise
Andreas Heil
, Johannes Meinecke
, Martin Gaedke
Chemnitz University of Technology
09111 Chemnitz, Germany
Microsoft Research Cambridge
CB3 0FB Cambridge, United Kingdom
Abstract: For a modern enterprise, it is vital to be on the Web. Beyond offering
human-readable Web sites, organizations increasingly use the Web as a media for
machine-readable data about itself. With the help of technologies like XML feeds,
RESTful Web services and semantic markup, new forms of enterprises models
emerge in a bottom-up way. These models are easily consumable and facilitate the
interaction with departments, partners and customers. Engineering good publishing
systems is however extremelychallenging. On the one hand, knowledge of many
technologies is required; on the other hand, it must be easy to extend systems and
data models in accordance to agile businesses. In this paper, we propose a frame-
work of components for publishing dynamically growing enterprise models on the
Web, present an implemented system and discuss its use in a case study.
1 Introduction
An enterprise model is a computational representation of an organization, covering as-
pects like its structure, activities or processes [FG98]. In recent years, there has been a
tendency for enterprise data to grow in a bottom-up way rather than in a planned top-
down way [TW06]. Simple lists, which were originally created to fulfil some personal
organization need, are made available to others, get extended and become valuable assets
for the enterprise. Technologically, this bottom-up form of collaboration is enabled
through Web-based tools like Wikis, blogs, feeds and mashups. These technologies are
suitable for both, in usage the Internet with public accessibility as well as usage in corpo-
rate networks and intranets. The dynamic growthof enterprise data in both scenarios is
favoured by the principles of the Web, after which relevant resources have a URI and
can thus be linked to and combined. In accordance to the Web’s underlying Representa-
tional State Transfer (REST) architecture [Fi00], companies with such Web-enabled
models could be called RESTful enterprises.
Thisway of exposing information comes with a number of advantagesfor the company.
The data can be easily combined in end user-written mashups that are characterized by
decreased implementation costs over traditional software development [An06]. Typical
scenarios involve the combinationof external sources, like geographical maps, with
internal data, like employee workplaces. Therelative simplicity of standards like HTML
and XML makes it easy to reuse the information in many places, as e.g. to display in-
formationonmultiple web sites. This in turn improves consistency and lowers the cost
to maintain the information, changing the future business behaviour [Yo07]. In addition,
existing Web tools can be leveraged, as for example to provide search across the com-
pany’s XML data sources [MM04].
In this paper, we concentrate on the question of how to build the necessary Web-based
systems to expose enterprisemodels efficiently, rather than on the enterprise models
themselves. In section 2, we examine the engineering challenges that arise in typical
scenarios with the help of an example. From this examination, we derive the aim of our
work: to identify reusablecomponents for exposing enterprise models in a Web-
compliant way. As our main contribution, we then specify a framework of components
that can be instantiated to support a large number of standards and tasks in section 3. In
section 4, we discuss our experience with the implementation of a number of these com-
ponents, which we usedto gradually build up a RESTfulenterprise model. Section 5
contrasts our work with related approaches and section 6 summarizes the conclusions.
2 Exposing Bottom-Up Enterprise Models on the Web
For illustration purposes, we explain the general problem domain with an example sce-
nario. The example covers the typical evolutionary bottom-up growth of information
assets within organizations that start within small teams and turn into large-scale under-
takings over time [TW06]. Figure 1 shows entities inside an organization that gradually
emerge as Web-based representations, called lists in the following.
Figure 1: Example of a gradually growing enterprise model exposed on the Web.
In the beginning, a team setsup lists as ad-hoc solutions for managing reports and meet-
ings. The content is published in the RSS format, which allows team members to sub-
scribe to automatic notifications on any new list items and which enables other teams to
integrate the lists in their homepages (1). Over time, this “model” is extended, as more
lists become available (2) and more data formats of the existing lists are offered (3). For
example, in addition to the very generic RSS format, other XML formats might include
geographical coordinates and domain-specific information that enable more powerful
mashups. In order to improve the combined use of the data, multiple lists are then linked
to each other (4). More precisely, URLs are included in the list items that point to other
list items, as e.g. to point from reports the authoring employees. Data from existing pro-
ject and customer management solutions is exposed in a similar way to be combined
with the ad-hoc data sources (5). Future interoperability needs may demand the support
for additional, semantically annotated formats (6). These can be searched, viewed and
processed with a wide range of semantic tools, without the needfor programming appli-
cation-specific mappings.
The described way of exposing enterprise data can be seen as an application of the REST
architecture style. REST comprises a set of design principles that are seen as the reason
for the scalability and organic growth of the Web[Fi00]. These principles are applied
here to achieve a similar advantage for enterprise models. A central requirement de-
mands that every important entity must have an identifier so that it can be referred to. As
shown in Figure 1, both lists and list items in the example are mapped to URIs. REST
distinguishes between resources (the entities on the left side) and their various represen-
tations (the content types on the right side). The representation useddepends on the
demands of the requesting application. In the example, the same resource is reused in
mashup engines, RSS readers, Web applications and Semantic Web browsers. All re-
sources can be manipulated in a standardized way through the uniform interface of
HTTP methods. In addition to read-access, this allows e.g. adding a new entry to the
meeting list via an HTTP POST.
Whereas tools now become available for easily combining data sources on the Web
[Ma08, Mi08, Ya07], exposing structured data beyond the simple publication of static
XML files remains a largely unsupported task. [PZL08] list several design decisions to
be made when programming RESTful services that include designing URIs, defining the
interaction semantics of HTTP methods and choosing the data presentations. In this
work we are interested in building systems that automate this process as much as possi-
ble to support the outlined target domain.We observe the following engineering chal-
Gap between end users and technologies: The initial data sources in the scenario
are set up by users rather than a central IT-department. This corresponds to the find-
ings of US Bureau of Labour Statistics [Us08], stating that 98% of users creating
programs are end users without special IT-expertise who need to solve immediate
non-technical problems. In contrast, the described final solution comprises a large
number of technologies. While the benefits of e.g. Semantic Web formats may be
highly desirable, the necessary knowledge to author them cannot be expected from
the end user. Instead, technologies should be encapsulated in tools or parts of tools as
much as possible.
Reusability of multiple resource representations: As exemplified, the reusability
of the exposed enterprise modelsrequires the same resources to be accessible in
many different representations. To free theuser from the burden of managing differ-
ent representations, these should be generated automatically. For automatic mappings
to work, structure must be imposed on thedata. While unstructured content, as e.g.
managed in Wikis, has the potential for organic growth, it cannot be reused easily
and is therefore lost for applications beyond human browsing.
Dynamic growth of data structures: The enterprise model in the example is gradu-
ally extended, as more data is added and linked to existing data. Corresponding sys-
tems must therefore support the effortless creation of new lists, including their struc-
ture and the mappings to their representations. Dynamic growth implies that this is
possible at runtime and on the Web. Ideally, the process of extending the model can
profit from the same simplicity and standard-conformance as used for manipulating
individual resources.
Need for system extensibility: In addition to the model, the systemitself is subject
to growth. In the example, this applies to the support for new formats (e.g. RDF),
new datasources (e.g. customer management systems) and new client applications
(e.g. search applications). Furthermore, the opening of themodel to partners outside
the enterprise may require adding security mechanisms that were unforeseen and un-
necessary at the very beginning. The system architecture should reflect this need and
allow for flexibility.
The observed challenges stress the need for solutions that comply with Web standards
and that encapsulate these standards in reusable, extendableparts. Next, we therefore try
to identify reusable components for exposing bottom-up enterprise models in aWeb-
compliant way.
3 The Data Grid Service Component Model
The WebComposition Data Grid Service (WebComposition/DGS) component model is a
two-layer framework of reusable software artefacts for building customizable Web-
based applications. A WebComposition/DGS component is a RESTful service for creat-
ing, managing and publishing lists of arbitrary data in the Web. It is based on a set of
exchangeablesub-components to incorporate the latest technological developments.
Exchanging these sub-components allows adopting the component itself to meet specific
business needs. In the followingwe will first describeWebComposition/DGS compo-
nent’s building blocks (cf. Figure 2) before we will discuss the possible usage of the
WebComposition/DGS component itself. Additional background information on the
WebComposition/DGS component and its architectural description can be found in
The Service Component (a) represents a central facet within the WebComposition/DGS
component model. Incoming requests are accepted, processed and delegated irrespec-
tively of their origination.Clients sending a request could be HTTP-basedbrowsers,
SOAP-based SOA components or legacy XML-RPC based clients that rely on so-called
plain old XML (POX) invocations. Incoming data is processed by the Data Adapter (b).
It is within the Data Adapter’s responsibility to create and store data structures in form
of lists using an external storage solution. Multiple Data Adapters could be specified
within a WebComposition/DGS component, each responsible for a specific representa-
tion of the resources. Based on the requestor’s needs, an appropriate Data Adapter is
used to provide the corresponding representation of a resource. In case of a HTTP-based
request this is achieved by HTTP content negotiation. Simultaneously to the creationof
the data, metadata is saved along with the data. Unnoticeable to the user, a Meta Store
component processes additional metadata (c). This includes e.g. creation and last update
times, creator and other information that describes the actual data. The metadata already
provided automatically by a Data Adapter can be completed by additional metadata
added through the user. As the format of metadata varies from case to case, Input and
Output Filter components (d) allow specifying the transformation of metadata formats
into and fromthe internal representationwithin the Meta Store component. To be de-
ployed in various business scenarios, an optional Access Control component incorpo-
rates the authentication and authorization functionality of external security components
(e). Finally, optional sub-components can be added to a WebComposition/DGS compo-
nent extending the Service Component’s capabilities with additional unforeseen func-
tionality (f).
Figure 2: UML component diagram of WebComposition/DGS sub-components.
The components described thus far allow customizing a WebComposition/DGS compo-
nent to a large extent. Once customized and configured, the WebComposition/DGS acts
as a component by itself that can to be seenas a black-box, hiding its internal functional-
ity (cf. Figure 3). A core concept of the WebComposition/DGS component (i) is its ca-
pability of being accessible in a uniform way through different types of clients. This
includes standard Web-based applications such as common Web sites (ii), RSS feed
readers as well as Semantic Web browsers. In addition, specialized clients can be used to
access more specific functionality of a WebComposition/DGS component (iii). One
example of this type of clients, a list managing application, will be discussed in the sec-
tion 4. Based on a generic SOAP interface, derived from our previous work [MMG07],
SOA-based clients can also access the WebComposition/DGS component (iv). The way
enterprise models are stored depends heavily on the particular requirements and con-
straints of a business. Hence, this is foreseen as self-subsistent component within the
system (v) allowing to apply database management systems or simple file based solution
based on the particular requirements. Thisalso allows reacting to the growing needs of
projects that originally start small and increase in size over time. Another form of poten-
tial evolution is foreseen by the security component that provides external, centralized
authentication and authorization functionality to complement the access control enforced
within the WebComposition/DGS component (vi). Possible security components applied
for this reason include, but are not limited to the Identity Federation System (idFS)
[MNG05], Shibboleth implementations [In08] or the Active Directory Federation Ser-
vices (ADFS) [Pi05].
Figure 3: Distributed application based on a WebComposition/DGS component.
The proposed two-layer component architecture allows the disciplined and cost efficient
engineering of Web-based application suitable to the current evolution stage. In the be-
ginning, the out-of-the box functionality of a WebComposition/DGS component in con-
junction with functionality of already available components could be sufficient to deal
with scenarios where unstructured data is created ad-hoc. The bottom-up growth of this
information as well as its increasingneedfor structures is tackled by exchangeable com-
ponents addressing the particular needs. In the following section we will discuss this
evolution of asystem, based on a real datastock that grew over time.
4 Case Study
We now describe a project in which the presented component model was applied to
bottom-up enterprise models. In the case at hand, the enterprise model represents the
structure of a research group. Performed in the academic field, the nature of the exposed
model is not restricted to the research domain, so that we consider the solutions applica-
ble to enterprises in general. The study was conducted for 6 months in a production
environment, usingreal, externally visible data. During this time, the model was gradu-
ally exposed and extended with new resources, new representations and new components
according to the emerging needs of the group. A part of it was transformed from origi-
nally unstructured media, mainly managed with Wiki software. Thisold form of data
management proved to be too hard to integrate and consume outside the Wiki itself.
Therefore, an implementation of the specified WebComposition/DGS was used to suc-
cessively expose data in accordance to Web standards and the REST principles. Over
time, this included publications, courses, projects, student projects and people. The lists
typically contain several hundred entries, describing both historical and currently active
data. In addition, the model was extended several times to accommodate for new infor-
mation needs (e.g. adding archive IDs to publications), to introduce linksbetween differ-
ent resources (e.g. relating publications and people) and to add new representations (cf.
Figure 4).
(a) (b)
(c) (d)
Figure 4: Representations of resources from the WebComposition/DGS in an RSS reader (a),in a
Semantic Web browser (b), in the DGS List Manager (c) and on a Web site (d).
The technical realization included the following components:
As the foundation, we used our standard implementation [HG08] of the WebCom-
position/DGS that is based on the .NET Framework. In this implementation, sub-
components (cf. Figure 2) correspond to dynamic link libraries that are composed ac-
cording to configuration files.
The XML Data Adapter provided the necessary facilities for managing arbitrary
XML lists. The adapter supports the creation of new lists at runtime as well as the en-
forcement of XML-Schemas to ensure that the data conforms to a given structure.
In addition to the default file-based XML storage, a Database Storage Solution
based on SQLite was developed to verify the exchangeability of storage solutions.
Another component, the Dynamic Transformation Adapter, was used to generate
new presentations from existing XML resources based on XSL stylesheets, as e.g. to
generate RSS-feeds (cf. Figure 4a).
Later, a RDF Data Adapter was developed to generate RDF representations of re-
sources and metadata according to Linked Data guidelines [BCH07]. Once a list
structure has been mapped to common ontologies, the model information can be re-
used in Semantic Web tools together with other data from the Web(cf. Figure 4b).
For managing the model stored in the WebComposition/DGS, a DGS ListManager
was developed. Thisgeneric client component retrieves the XML schemas from the
list manager and automatically generates forms for editing resources (cf. Figure 4c).
The integration into the research group’s Web site was supported with a PHP List
Client Component. The component encapsulates the technical details of rendering
the dataand generating the necessary links to other representations (cf. Figure 4d).
We summarize the lessons learned during the case study with respect to the engineering
challenges outlined in section 2. The concept of encapsulating technologies in compo-
nents proofed to be an important factor for end user support. With the developed system,
new lists can be created ad-hoc and are automatically editable in Web forms, without
any scripting or code deployment. They can also be integrated into Web pages without
the need to know the involved internal components, transformations, protocols and for-
mats. Whereas in our current system XML schemas and XSL stylesheets need to be
specified when creating new lists, the architecture allows for adding more user-friendly
components that automate this process further. The applied components also favored the
reusability of the resources by automating the process of generating content representa-
tions. On the Web site alone, the data could be integrated at multiple locations for realiz-
ing different views on it, e.g. on personal homepages, on project pages and on central
group pages. Furthermore, the study illustrated the system’s potential for bottom-up data
growth. Natural limits to flexible changes arose from theWeb’s need for stable URIs
and from necessary updates of the corresponding mappings into formats and ontologies.
The demand for system extensibility was met by the component-based architecture.As
demonstrated, components were gradually added to the WebComposition/DGS, while
the service was in productive use, i.e. integrated into the group’s Web site.
The described systems are still in use and will be extended in future for further studies.
A demo of the WebComposition/DGS as well as downloadable components can be
found at
5 Related Work
The growing demand for creating, managing and publishing data within Web-based
solutions is reflected in the high number of ongoing developments in this field. In this
section we discuss approaches related to our work. We concentrate on solutions that
follow the REST-style principles, allow building and hosting enterprise models and are
applicable to our case study. The solutions are segmented into application-specific, data-
base-driven and protocol-driven approaches.
Application-specific container solutions: This class of solutions includes Wikis, We-
blogs andportals that maintain data in the form of lists. Wikis provide lists with exten-
sive change-histories and references to list items that do not yet exist. Weblogs are also
part of this class by providing chronologically ordered lists of entries and different views
(e.g. by tags, month, week or day) on the data. While both follow REST-like architec-
tures and gain increasing recognitionwithin companies [Li07], they mainly focus un-
structuredcontent for human readers. An example of a commercial list-oriented applica-
tion is Backpack [37Si08]. This easy-to-useWeb-based service provides basic function-
ality to manage and share lists such as notes, images and files on the Web. The data
model is however preset andnot subject to growth. The more complex document man-
agement systemMicrosoft Office SharePoint [Mi07] provides the capability of maintain-
ing lists of documents, tasks, contacts, appointments as well as self-defined lists. While
this theoretically allows managing arbitrary models, the monolithic approach taken
makes it hard to reuse the structure outside the portal itself in a standardized way. Gen-
erally, container solutions are suitable for dedicated tasks, offer a limited set of function-
ality and appear mostly as stand-alone solutions that are difficult to integrate into exist-
ing component-based environments.
Database-driven approaches: Currently, anumber of services can be observed that
provide traditional database functionality over Web standards. These services mostly
target a commercial audience and are offered by particular providers. The Amazon Sim-
pleDB [Am08] Web service includes a simple REST-style Web service interface that
follows the traditional CRUD (Create, Read, Delete, Update) pattern [Ki90]. SimpleDB
is designed to store relatively small amounts of data with the focus on fast data access.
For growing needs, the commercial Amazon S3 service [Sh07] is offered. In contrast to
WebComposition/DGS, these services form black-box components, lack the capability
of customization and do not impose structure on the data. The Microsoft ADO.NET Data
Services [Ca07] comprise an implemented set of patterns for data-centric services. The
data is represented in various common formats such as XML, JSON or ATOM/RSS. All
the characteristics of the REST-like architectural style appear as proposed in [Fi00].
Compared to the WebComposition/DGS component, this solution is designed for the
Microsoft database management systems only and introduces dependencies to very spe-
cific platforms and technologies. None of the examined database-driven approaches
supported creating and changing list structures via their Web interface. Their usage often
requires extensive backgroundknowledge of database management systems.
Protocol-driven approaches: The Atom standard [GD07, NS05] defines an XML-based
format and an HTTP-based protocol for publishing and editing lists of related docu-
ments. For example, the POSTmethod is used to create new resources within a collec-
tion, including an additional link entry for an extendable set of metadata about the par-
ticular resource. The most common usage of Atom is the syndication of news and We-
blogs. While the format is extendable, there is no support for serving multiple represen-
tations of a resource. The Google Data APIs [Go08] provide simple protocols for reading
andwriting data on the Web based on RSS 2.0 andATOM 1.0. The CRUD concept is
fully supported through a REST-style interface using HTTP and the corresponding
methods to access and query the XML-based data. Metadata is provided in the form of
additional feeds containing referential information using e.g. the Google Base schema.
However, this strategy is limited in terms of the fixed semantic information provided by
the metadata feeds. According to their protocol-driven nature, the approaches do not
cover component-based architectures for realizing extendable systems.
With respect to the aimed usage for bottom-up enterprise models, the related approaches
were either too rigid for gradual growth, or toounstructured forproviding the model data
in multiple reusable representations.
6 Conclusion
In this paper we presented the WebComposition/DGS component model as a framework
for the Web-compliant expositionof enterprise models. Our work focuses on the often
observed bottom-up approach to creating valuable enterprise data sources that start out
as simple lists on the Web. This process is supported by breaking down functionality and
technology support into an extendable set of components. Within a case study, we
showed the practical applicationof corresponding component implementations and
demonstrated their ability to represent different representations of the model for maxi-
mum reusability. Upcoming work is targeted at finding ways to better support enduser
tasks, like e.g. the specificationof new lists. Another interestingopenproblem is the
transparent handling of URIs as references to other list entries or machine-readable in-
formation on the Web with corresponding user interfaces.
[37Si08] 37signals LLC: Backpack - Website, 2008, (02-19-2008).
[Am08] Amazon Web Services LLC: Amazon SimpleDB Developer Guide, 2008.
[An06] Anant, J.: Enterprise Information Mashups: Integrating Information, Simply, in 32nd
International Conference on VeryLarge Data Bases, VLDB Endowment, Seoul, Korea;
2006, p. 3-4.
[BCH07] Bizer, C.; Cyganiak, R.; Heath, T.: How to Publish Linked Data on the Web, 2007.
[Ca07] Castro, P.: Project Astoria, The Architecture Journal, 2007(13): p. 12-17.
[Fi00] Fielding, R.T.: Architectural Styles and the Design of Network-based Software Architec-
tures, University of California, 2000.
[FG98] Fox, M.S.; Gruninger, M.: Enterprise Modeling, AI Magazine, 1998. 19(3): p. 109-121.
[Go08] Google Inc.: Google Data APIs - Website, 2008, (02-
[GD07] Gregorio, J.; Hora, B.D.: The Atom Publishing Protocol - Request for Comments: 5023,
2007, (02-19-2008).
[HG08] Heil, A.; Gaedke, M.: WebComposition/DGS: Supporting Web2.0 Developments With
Data Grids, in IEEE International Conference on Web Services (ICWS 2008), Beijing,
China; 2008.
[In08] Internet2: Shibboleth - Website, 2008, (08-08-2008).
[Ki90] Kilov, H.: From semantic to Object-oriented Data Modeling, in First international Con-
ference on Systems Integration,Morristown, NJ, USA; 1990, p. 385-393.
[Li07] Li, C.; Stromberg, C.: The ROI of Blogging, Forrester Research Inc., 2007.
[Ma08] Markl, V. et al.: Data Mashups for Situational Applications, in Model-Based Software
and Data Integration: First International Workshop, Mbsdi 2008,, Berlin, Germany;
[MMG07]Meinecke, J.; Majer, F.; Gaedke, M.: Component-Based Content Linking Beyond the
Application in Seventh International Conference on Web Engineering (ICWE), Springer,
Como, Italy; 2007, p. 427-441.
[MNG05]Meinecke, J.; Nussbaumer, M.; Gaedke, M.: Building Blocks for Identity Federations, in
Fifth International Conference for Web Engineering (ICWE2005), Springer, Sydney,
Australia; 2005, p. 203-208.
[Mi07] Microsoft: Microsoft Office SharePoint Server 2007 Homepage - Website, 2007, (02-01-2007).
[Mi08] Microsoft Corporation: Microsoft Popfly - Website, 2008, (02-
[MM04] Mukherjee, R.; Mao, J.: Enterprise Search: Tough Stuff, Queue, 2004. 2(2): p. 36-46.
[NS05] Nottingham, M.; Sayre, R.: The Atom Syndication Format - Request for Comments:
4287, 2005, (02-18-2008).
[PZL08] Pautasso, C.; Zimmermann, O.; Leymann, F.: RESTful Web Services vs. “Big” Web
Services: Making the Right Architectural Decision, in 17th International World Wide
Web Conference, ACM Press, Beijing, China; 2008, p. 805-814.
[Pi05] Pierson, N.: Overview of Active Directory Federation Services in Windows Server 2003
R2, 2005.
[Sh07] Shanahan, F.: MashupsEds., Birmingham, UK: Wrox Press Ltd., 2007.
[TW06] Tapscott, D.; Williams, A.D.: Wikinomics: How Mass Collaboration Changes Every-
thing, New York, NY: B&T, 2006.
[Us08] U.S. Department of Labor: Bureau of Labor Statistics - Website, 2008, (07-20-2008).
[Ya07] Yahoo! Inc.: Pipes: Rewire the web - Website, 2007, (02-
[Yo07] Young, G.O.: IT Will Measure Web 2.0 Tools Like Any Other App, Forrester Research,
Inc., Cambridge, MA, USA, 2007.
Conference Paper
Full-text available
Recent technology trends in the Web Services (WS) domain in- dicate that a solution eliminating the presumed complexity of the WS-* standards may be in sight: advocates of REpresentational State Transfer (REST) have come to believe that their ideas ex- plaining why the World Wide Web works are just as applicable to solve enterprise application integration problems and to simplify the plumbing required to build service-oriented architectures. In this paper we objectify the WS-* vs. REST debate by giving a quantitative technical comparison based on architectural principles and decisions. We show that the two approaches differ in the num- ber of architectural decisions that must be made and in the number of available alternatives. This discrepancy between freedom-from- choice and freedom-of-choice explains the complexity difference perceived. However, we also show that there are significant dif- ferences in the consequences of certain decisions in terms of re- sulting development and maintenance costs. Our comparison helps technical decision makers to assess the two integration styles and technologies more objectively and select the one that best fits their needs: REST is well suited for basic, ad hoc integration scenarios, WS-* is more flexible and addresses advanced quality of service requirements commonly occurring in enterprise computing.
Conference Paper
Full-text available
Technologies like XML and Web Services have posed new requirements to authentication, authorization and identity management for the Web as an application platform. Beyond merely providing access control for a single isolated system, modern, flexible architectures support a business-spanning federation of applications and services by sharing digital identities. The diversity of today’s specifications and the many aspects to be considered, like e.g. privacy, system integrity and distribution in the Web, makes the construction of these architectures a very time-consuming task. Thus, a uniform view on the overall system is needed that abstracts from technological issues. This can be achieved by extracting the core concepts from the emerging Federation technologies and specifications and formalize them to an extent that they can be used as a foundation for configurable applications and services. In this paper, we introduce a solution catalogue of reusable building blocks for Identity and Access Management (IAM). We also present a configurable system that supports IAM solutions in Web-service-based applications.
Conference Paper
Full-text available
The content of many innovative Web sites today often originates from beyond the application. This paper is concerned with building Web appli- cations that heavily integrate and link content from external sources, like e.g. Web services or RSS feeds. Unlike conventional applications, they are charac- terized by a very dynamic and distributed information space. In this context, traditional Web Engineering approaches suffer from the fact that they rely too much on a-priori knowledge of existing content structures. We present a sup- port system and a method for building such applications in a very flexible way. Flexibility is achieved by managing links separately from the content in a dedi- cated Web service and by composing the application from fine-grained, reus- able components that realize navigation, presentation, and interaction for the linked content.
Conference Paper
The need to create, store, maintain, and share data as being part of Web 2.0 labelled solutions requires the application of different skills and tools sets. The growing complexity of these new kinds of applications leads to cost and time-intensive development life-cycle because many data design and management decisions are directly influenced by distribution and semantic aspects. In this paper we propose the Web composition data grid service (Web Composition/DGS) as a supporting service to address these problems. The service acts as a highly integrative core component for data-centric applications, taking traditional SOA-based business scenarios as well as REST-architectural style applications in the Web 2.0 context into account.
Conference Paper
There is a fundamental transformation that is taking place on the web around information composition through mashups. We first describe this transformation and then assert that this will also affect enterprise architectures. Currently the state-of-the-art in enterprises around information composition is federation and other integration technologies. These scale well, and are well worth the upfront investment for enterprise class, long-lived applications. However, there are many information composition tasks that are not currently well served by these architectures. The needs of Situational Applications (i. e. applications that come together for solving some immediate business problems) are one such set of tasks. Augmenting structured data with unstructured information is another such task. Our hypothesis is that a new class of integration technologies will emerge to serve these tasks, and we call it an enterprise information mashup fabric. In the talk, we discuss the information management primitives that are needed for this fabric, the various options that exist for implementation, and pose several, currently unanswered, research questions.
The last decade has witnessed the growth of information retrieval from a boutique discipline in information and library science to an everyday experience for billions of people around the world. This revolution has been driven in large measure by the Internet, with vendors focused on search and navigation of Web resources and Web content management. Simultaneously, enterprises have invested in networking all of their information together to the point where it is increasingly possible for employees to have a single window into the enterprise. Although these employees seek Web-like experiences in the enterprise, the Internet and enterprise domains differ fundamentally in the nature of the content, user behavior, and economic motivations.
Conference Paper
The author shows a possible path from semantic data modeling and data manipulation to greater support of information integrity by means of applying programming methodology paradigms, i.e. object-oriented data modeling and data manipulation. It is suggested that the dissimilarity between the semantic and the object-oriented models is not as sharp as it may seem. The trend for current experimental and commercial database models is to have both facilities: both structural and behavioral properties of data should belong to the database schema, i.e. to a data definition language. Extensibility of the sets of available metatypes and of available types and explicit support of high-level behavior at the metatype and type levels lead to the integration of these two models. It is therefore concluded that the path from an extensible semantic model to an object-oriented one is within the reach of the current generation of data modelers and users