aa r X i v : . [ c s . D C ] S e p Business in the Grid
Erich Schikuta
Thomas Weish¨aupl
Flavia Donno + Heinz Stockinger ∗ Elisabeth Vinek + Helmut Wanek
Christoph Witzany
Irfan Ul Haq
University of Vienna, Department of Knowledge and Business EngineeringRathausstraße 19/9, A-1010 Vienna, Austria ∗ Swiss Institute of Bioinformatics, CH-1015 Lausanne, Switzerland + CERN, CH-1211 Geneve 23, Switzerland
Abstract
From 2004 to 2007 the Business In the Grid (BIG) project took place and was drivenby the following goals: Firstly, make business aware of Grid technology and, secondly,try to explore new business models. We disseminated Grid computing by mainly con-centrating on the central European market and interviewed several companies in orderto gain insights into the Grid acceptance in industrial environments. In this article wepresent the results of the project, elaborate on a critical discussion on business adapta-tions, and describe a novel dynamic authorization workflow for business processes inthe Grid.
Grid computing finds its origin in the academic world where it was “created” and hasgained a lot of popularity in the last decade. Since then, many national and internatio-nal Grid oriented solutions have been devised specially to harness huge data and solvecomputationally intensive problems. A few years ago, commercial partners also joinedGrid efforts in order to profit from this promising technology. However, the Grid com-munity is still in search of the so called “killer application” and is experimenting withapplication scenarios where the commercial world can really make profit by utilisingGrid technologies.Nowadays it can be learned from the developments in the IT market that there isan obvious shift in the source of profits, from sales of products to the provision of on-demand services [41]. This leads to the development of novel business models wherenew application domains will share applications and resources. Thus for companies inthe IT market to be successful the availability of techniques allowing for integrationand collaboration of resources of any kind is a crucial issue.1he Grid promises the revolution of the Internet by a novel and advanced supportfor collaboration providing homogeneous access to virtual resources without revealingthe heterogeneous manner of the underlying real world. Basically the Grid resemblesa distributed computing model supporting the selection, sharing, and aggregation ofgeographically distributed ”autonomous“ resources dynamically at runtime dependingon their availability, capabilities, performance, costs, and users’ quality-of service re-quirements via Grid-enabled Web services [30].In the literature a large number of different definitions of the notion of Grid can befound [50]. This is connected with an ongoing discussion within the research commu-nity what the specific characteristics of a Grid really are and what makes it so differentfrom the Internet and the Web. From our business oriented point of view we identifiedthree necessary specifics of the Grid infrastructure to allow for business workflows,which are: • Transparency.
The parties involved in sharing of resources are anonymous toeach other, that means the consumer of a resource does not need any knowledgeabout the provider of the resource and vice versa. • Generalized Quality of Service.
QoS is more than technical properties. It is bet-ter described by the notion of SLA (Service Level Agreement) and has to compri-se all necessary aspects of business resulting in a ”trust relationship“ between cu-stomer and provider. This can be confidentiality, data integrity, non-repudiation,accountability, . . . • Brokerage.
Resources are made available by a Broker, which is in our termi-nology a policy-based mediator of services and business workflows aiming formaximizing profit of all partners.In our vision it will be possible to sell software and resources as a service and notas a good in the near future.For example,
Writing a letter can be as simple as using a telephone: Forget buyingsoftware and hardware! All we need is a simple interface to the services on the Grid,which could be a cell phone. Both the wordprocessor functionality and the necessaryphysical resources (processor cycles and storage space) are available as services on theGrid; and we pay just for the usage transparently via our telephone bill.Until now the research community mainly focused on the technical aspects of Gridcomputing neglecting commercial issues. However now there is a focus shift towardsthe commercial exploitation of Grid computing.In 2004 we created the project Business in the Grid (BIG) in order to analyse theIT market and its current acceptance to Grid computing in business environments. Wemainly focused on the central European market but believe that the results are repre-sentative for other countries, too.Preliminary results of the projects have already been published earlier ([48]). In thisarticle, we provide more in-depth and updated information on our findings. The articleis organised as follows. We will first review the project goals in Section 2 and thenrevise existing business models. These models are then used to analyse the IT marketby first proposing a dictionary (Section 4) and then a detailed market analysis along2ith an on-line survey. Finally, we discuss new business models and present as novelscientific result of the project the gSET method for dynamic authorization in businessworkflows on the Grid.
The project Business In the Grid (BIG) was created with the purpose to analyse the ITmarket in order to find out if there is a market potential to use Grid technologies in acommercial environment. In detail, the project had the following goals, as describedin [57]:1.
Revisiting of existing (E-)business models for the Grid.
Existing (E-)business models were revisited for a possible adaptation to the Grid.Reasons for the failure of existing business models for E-Commerce in the In-ternet and Business to Business (B2B) were analysed. It was observed that theGrid can provide chances of success for these models by new transparent layers (included security mechanisms, support of dynamic services, etc.).2.
Market potential analysis and information dissemination.
Information dissemination and market potential and demand analysis was mainlyfocused on the Austrian economy landscape, but some other countries such asGermany, Czech Republic and USA were also kept in view.3.
Development of novel business models for the Grid.
The main goal of the BIG project was to find novel business models enabled bythe Grid infrastructure. New ways of doing business in the Grid were discover-ed. It was realized that there are possibilities for dynamic collaborations, newproject workflows, software on demand, dynamic resource-management, resour-ce on demand, application service providers, and other Grid information societycomponents.
In the first phase of the BIG project, a broad state of the art analysis was done thatrevealed very interesting information. It was noticed that no standard terminology isused in “economy Grid” projects in the community. Until now no clear understandingand categorization exists in this area. Problem formulations of many “economy Grid”projects have difficult lingual descriptions, based on vague assumptions thus resultingin weak problem and requirement specifications.The state of the art analysis shows that economic aspects influence Grid computingprojects on different layers. The terms business, commerce, and economy are not usedconsistently. This makes it difficult to proceed efficiently to the commercialization ofvirtual resources. 3 rid Middleware - TO SOLVE HUGE PROBLEMS-BY SHARING
Economy Principles
IN MIDDLEWARE-Negotiation-Guarantees-Quality of Services-Auction etc.
Business Enabled
Middleware-Trading-Accounting-Payment etc.
E-COMMERCE -Security-Usability-Legal environment-Business models-Customer locality(trust, …)MarketGoodsInfrastructure BusinessProblemSolution
Abbildung 1: Grid development to businessThe purpose of the following chapter is to clarify the relation between Grid com-puting and economy with the final goal to transform virtual resources to commodities.We proposed a layered architecture [57] to describe different economic aspects of theGrid. The model is called Economy-Grid layer (EG layer) model.
The EG layer model is derived from the background of historical observation. Differenthypotheses and theories [23] exist which aim to explain the reasons for the technologi-cal developments in relation to economy and innovation.We use simplified categories describing the history of the development of existingtechnologies. The categories result from a “problem-solution” process with the respec-tive steps: problem, solution, good, infrastructure, market, business. Figure 1 visualizesthe mentioned process by arrows below the cycles. We show the validity of the simpli-fied process by two examples:Telephony is a representative technology, which has followed this historical process[13, 24]. At the beginning the transmission of voice signals between two dislocatedpoints was achieved (problem-solution). It was not trivial that this solution developedinto a good. It took many years from the discovery of the telephone to the good, whichmeans that somebody would buy and pay for the service or good. An infrastructure hadto be built to deal with the good. Accounting and other systems were necessary. By theinfrastructure it was possible that a market with customers could be established. Initial-ly, the telecom markets were quite small, homogeneous, and limited (e.g. one providerper country or region). Therefore, we distinguish a further step, called business, whichrepresents a multi provider market.As another example for this process the evolution and usage of the Internet can4e mentioned. Considering the Internet, which was invented to solve communicationchallenges in the area of military defence, is now a communication infrastructure foreverybody, used also for doing commerce (e.g. Amazon, EBay etc.). Obviously out ofa problem and its solution arises a new good. This good can be traded in an adaptedinfrastructure, which creates a market and therefore a new commercial business.The cycles of Figure 1 visualize a possible evolution of Grid technology. Grid midd-leware was developed to solve huge computational problems by sharing of resourcesinside a virtual organization (see [22]). The Grid middleware uses not only principlesand concepts of computer science, but also economy to provide the capabilities mentio-ned above, e.g. [12]. This middleware is a new commodity or product. Until now, thedevelopments are in progress and no finalized “business enabled” middleware exists(see [27]), which can establish a new market or field of commerce.By the allocation of the Grid development to the historical process of commontechnologies we presume the future development and derive the EG layer model. Allprojects described in the next section can be categorized in this way. However, this listshould not be taken as exhaustive.
The model results from the observations and conclusions mentioned above. We proposethe Economy-Grid layer (EG layer) model, to better understand the problems in contextof using a Grid infrastructure as a new good or product.The EG-layer model consists of four layers with the following characteristics:
EG1 Integration Layer: Grid using economic principles
Economic principles, concepts, and experience are integrated into the Grid andinfluence developments of Grid infrastructure, e.g. resource usage can be optimi-zed by the adaptation of auction principles. Typical representatives are the eco-nomy based replica optimizer [12] and the Gridbus project [29] with the GRACE(Nimrod-G) component [15]. The GEMSS project [25] and the NeuroWeb pro-ject [47] are also examples for EG1, which negotiate quality of services (QoS)between clients and service providers.
EG2 Commercialization Layer: Selling Grid software
Companies create products or services by using Grid software or some Gridcomponents for “homogeneous” organizations. They sell the recent open sourcesoftware combined with self-developed software modules and services. Repre-sentatives of EG2 are for example DataSynapse [17], Avaki [52], Univa [54] orOrcale.
EG3 Enabling Layer: Business enabled Grid
The business enabled Grid establishes an open Grid, with similar properties asthe Internet for information today. An infrastructure for a market has to be pro-vided. In a Grid the resources can not be free, but accessible under user cons-traints. A market can regulate resource sharing in a satisfying way for a resourceprovider and customer.
Single companies need trading, accounting and payment5 c ono m y – B u s i ne ss T r ad i ng A cc oun t i ng P a y m en t Time
Business models on GridGrid using economic principles Selling Grid softwareBusiness enabled Grid
Griddevelopment recent research dependen cy (cid:1)(cid:0)(cid:2) (cid:3)(cid:4)(cid:5) (cid:6)(cid:7)(cid:8) (cid:9)(cid:10)(cid:11) Abbildung 2: Economy-Grid layer (EG layer) modelmechanisms. Representative components of EG3 are the BEinGrid [2], Adapti-ve Service Grid (ASG) [1], CHALLENGERS [5], Financial Service Grid (Fin-GRID) [6], BREIN [4], GridBank [11], GRIA [28], an OGSA-Based AccountingSystem [46], and the GGF GESA-WG [26].
EG4 Modelling Layer: Business Models on Grid
The market enabled Grid infrastructure gives possibilities for new business mo-dels and
E-Commerce . Implementation and any significant work on Layer EG4is not done until now. No open real markets exist for virtual resources. The lackof rentability and real business use-cases block further developments. Nonethe-less, there are some projects such as GridEcon [7], BIS-GRID (2007-2010) [3]and NESSIGrid [8], which have vowed to work on business models for the Grid.The BIG [55] project has also worked on this layer.Figure 2 shows the context of the EG-layer model with Grid development progress,time and economic usage. The recent research work is done on the first three layersonly. Higher layer depends on lower layer, whereas the width of the light-gray verticalbars represents the quantity of dependency. Layer 1 interacts with Layer 3, because e.g.about a price of a resource of Layer 3 has to be agreed by a resource allocation mecha-nism (broker, scheduler) of Layer 1. All layers are necessary to establish the economicusage of a future Grid. The difference between Layer 3 and Layer 4 is, that Layer 3infrastructure provides business between known partners. The Layer 4 infrastructureestablishes sophisticated business models and highly dynamic virtual organizations. Abusiness model describes the operations of a business including the components of thebusiness, the functions of the business, and the revenues and expenses that the businessgenerates. 6
Dictionary on Business in the Grid
In order to provide an overview about different Grid and business activities, we havecreated a dictionary. It was written on a Web platform using a special WIKI technology,which is particular advantageous for a dictionary because of the many cross links thatare inserted automatically.In order to keep it concise, the items constituting the dictionary were divided intofive sections:
Projects , Conferences , Companies , Products , and
Dictionary that consti-tutes the core part. In the following the most interesting facts and the main results ofeach section are described.
There are already many projects in the area of Grid computing not only dealing withtechnical issues but also with issues concerning business aspects of the Grid. The In-formation Society Technology (IST) priority program from the European Union - oneof the priorities of both the fifth (1998-2002) and the sixth (2002-2006) FrameworkProgrammes (FP5, FP6) from the European Union - has a sub-unit called Grid Techno-logies [34] that comprises EU-funded RTD activities in the area of ‘Grid-based systemsfor Complex Problem Solving’. This topic been chosen as one of the ‘Strategic Objec-tives’ of IST under the specific program ‘Integrating and Strengthening the EuropeanResearch Area’ of FP6. Some of the Grid-related projects deal explicitly with businessaspects of the Grid, as for instance the GRIA project [28] under FP5 and InteliGrid[33], NextGRID [42] or Trustcom [53] under FP6. And of course there are several in-ternational projects focusing on the same aspects, as for instance the Japanese BizGridproject [14], the Australian projects Economy Grid [18] and Gridbus [29], the UK GridMarkets Project [40] or the Grid Economics project [31] hosted at the Zurich IBM Lab.Most of the projects have partners both from academia and industry. Some of themproduce output that is directly used by the industrial partners. This is for instance thecase in the Corporate Ontology Grid (COG) project [16]: the results obtained during theproject were directly applied to the automotive industry. As in numerous other projects,one of the goals was to demonstrate the applicability of Grid technologies to industry.The projects can roughly be classified in two categories: there are software-centredprojects and architecture-centred projects. The software-centred projects can mainly belocated on the EG1 - Integration layer of the Economy-Grid layer model [59] presentedearlier. The outputs of these projects are mainly Grid-related software packages that useeconomic principles, such as accounting systems and implementations of negotiationmodels between the resources. As examples we can mention the SWEGRID project[51] and the Gridbus project [29]. The SGAS (Swegrid Accounting System) developedwithin the SWEGRID project is a Java implementation of a Grid Accounting Systembased on the OGF Open Grid Service Architecture. The middleware developed withinthe Gridbus project is engaged to support e-science and e-business applications. Amongother features it includes a competitive economy-based Grid scheduler, a Web-servicesbased Grid market directory and Grid accounting services. Architecture-centred pro-jects are on a more abstract basis (but of course some of them do also develop somesoftware) and can mainly be located on the EG3 - Enabling layer of the Economy-Grid7ayer model [59], for instance the GRIA project [28]. GRIA is a Grid aimed at businessusers and enables commercial use of the Grid in a secure, interoperable and flexiblemanner. The goals are to allow industrial users to trade computational resources on acommercial basis to meet their needs more cost-effectively and to allow service provi-ders to rent out spare CPU cycles and thus to allow clients to hire those CPU cycles. Inthe longer term, the project partners aim to use GRIA as a vehicle for prototyping newbusiness models on the Grid.There are several issues appearing in mostly all projects. The most important oneseems to be security. In fact, one big challenge in moving from academic Grid pro-totypes to enterprise Grids is the need for a high security level in the enterprise envi-ronment. This issue becomes even more important when the Grid infrastructure jumpsout of the boundaries of a single enterprise. Moreover, nearly all projects are commit-ted to contribute to the development of standards, which are a main requirement ofthe industry to assure interoperability among various systems. Another important pointis simplicity: in the community there seems to be the conviction that only relativelysimple systems can survive in the market.Until now, there have been only very few projects that could be located on theEG4 - Modelling layer of the Economy-Grid layer model [59], and much work needsto be done at this level to make the economy prepared for the rapidly evolving Gridinfrastructure and many possibilities within it. Until now, there seems to be a lack ofclear definitions and market-regulating mechanisms, as for instance pricing models.
Using the Grid for developing new business models is still a rather new idea, but thereare already some organizations putting a lot of effort in this area. The two key orga-nizations were the Global Grid Forum and the Enterprise Grid Alliance [19]; the mostrelevant conference is the GridWorld conference that took place for the first time inautumn 2005. The Global Grid Forum (GGF) and the Enterprise Grid Alliance (EGA)have merged together on June 26, 2006 to form Open Grid Forum (OGF).
During the past five years, many companies providing Grid or Grid-related productsand services appeared on the market, often as spin-offs from universities. We tried toclassify these companies following the Economy-Grid layer model [59] presented ear-lier in this chapter. It appears that most of the companies can be located on the EG2 -Commercialisation layer: they are mostly selling Grid software. As examples, we cancite Avaki [52], Axceleon [10], GridSystems [32] or United Devices, and also muchlarger and established companies such as Oracle and IBM. However, some companiesare getting a step further and can already be located on the EG3 - Enabling layer, asParabon Inc. [44] for instance. Parabon provides a compute engine for free downloadand thus allows everybody to become a paid provider. Jobs submitted by paying cli-ents are split in small tasks and distributed among the providers that perform them intheir computer’s idle time, without even noticing it. The results are then reassembledand sent back to the client. The price calculation is among other things based on task8rioritization: the more quickly a client wants his job to be executed, the more he hasto pay. However, most of the time the companies implementing such a system also sellsome software components and it is not clear to what extent the usage of a real Gridinfrastructure for doing business contributes to the overall revenue.
We summarized different topics and concepts concerning business in the Grid and notfitting in any of the other categories. These topics include architectural issues suchas Service Oriented Architecture and the Open Grid Service Architecture (OGSA),concepts and technologies enabling business in the Grid, such as Virtualization, WebServices and Grid Services, models such as the Economy-Grid layer model [59] andthe Enterprise Grid Alliance Reference Model [9], standardisation issues and concretestandards such as WS-Agreement, and topics of general interest for business in theGrid such as Grid Security, Grid Pricing Models, Grid Accounting, etc. Some of theseissues are already mature, as Grid security for instance: because of the strong need fora secure environment in enterprise Grids, much effort has been put on security issuesand technology seems to be very advanced in this area. On the other hand, some of thementioned topics are only at the very beginning of the development. This is for instancethe case with the pricing models in the context of business Grids. There is not muchliterature on this topic yet and there are no concrete proposals on how to regulate thepricing for resources in a commercial Grid environment.A software licensing model is one of the things often discussed when using com-mercial software on the Grid but no consensus has been reached on that topic. Howe-ver, it is clearly visible that the commercial usage of Grids or more generally speakingbusiness in the Grid is driving forward. More and more companies are adopting Gridtechnologies both inside their business and for cooperating with partners. One indicati-on for this development is the Oracle Grid Index Report [45] performed by the OracleCorporation and the IT analyst company Quocirca. The report aims at analysing com-panies’ attitudes towards the adoption of Grid-related technologies.9n conclusion, we can say that there are different levels of doing business in theGrid - or different ways to interpret the expression “business in the Grid”. These levelsare conforming to the layers of the Economy-Grid layer model [59] developed wi-thin the BIG project. The dictionary gives an overview of projects, alliances, products,companies and concepts located on any of these levels. The challenge about the Gridand especially about doing business in the Grid infrastructure is not so much techni-cal but to make work together people from IT management, computer scientists, otherscientists and engineers, vendors and people from the commercial world, that have alldifferent visions and expectations towards the Grid. • Is there really a commercial market for the Grid? • How do business and IT leaders think about (adopting) Grid technologies?Many people in the Grid community have their personal opinions on these que-stions, and they are very diverse from being very positive to very negative. However,we took the approach that we directly contacted several national (Austrian) and multi-national companies in order to interview them face-to-face about the market potentialsof Grids. In total, we interviewed more than 25 key persons from different companiesand organizations, in order to get an idea about the current perception of Grid tech-nologies in business and industry. Interview partners were mainly the company leaders(CEOs), IT leaders (CIOs, IT managers and IT project managers) as well as technologyleaders and people involved in education. Therefore, we gained a good overview aboutcurrent strategic decisions of companies with respect to Grid computing. Details on theinterview series are given in Section 5.1.We also started a second survey that potentially reaches a bigger community ofcompanies since the interview forms are on-line and no direct interviews are necessa-ry. This interview series was done in co-operation with the CAS Community [35], thathelped us promote the Web site to a large number of companies world-wide, mainly inthe storage business. A detailed discussion on the on-line survey is given in Section 5.2.As a short summary, we can deduce the following: • Only a rather small number of companies are currently actively deploying Grids or related technologies. Often, these are multi-national companies. • For small and medium companies (SME), the idea of
Grid computing is not yetdiscussed much since even technologies such as Web services are yet not fullyadopted. 10 ifferent market sectors.
ITEducationGovernment relatedIndustry (Auto, Steel,Chem.)InsuranceLogistics
Abbildung 3: Different market sectors represented in the interview series. Some com-panies or representatives operate in two sectors (e.g. IT and Education) • All companies see Grid computing as a way to reduce costs in certain areas, eit-her for providing new services to customers (although such services are current-ly very rare) or by outsourcing customer related services or computing servicesthat are related to the production chain of companies. However, outsourcing isconsidered to be more attractive for bigger rather than smaller companies. • Some companies (such as ASPs) state that they make more money with sellinglicenses rather than selling services.In the following two sections we go into the details of the personal interviews andthe on-line survey.
The usage of Grid technology in an industrial environment or within a company canbe regarded as a strategic decision such as the adoption of a company wide Intranetinfrastructure. Consequently, in order to get Grid technologies accepted within suchdomains, it is important to first “convince” representatives of strategic managementor opinion leaders and then people at the operational level. We therefore created aninterview series that was particularly focused on opinion leaders.The main idea was to personally talk to people who represent companies of diffe-rent business sectors (see Figure 3). We either contacted people directly via e-mail orphone, or filled in requests in company Web sites that then referred us to the respectivepeople. In general, about 70% of the contacted companies or representatives welcomedus for interviews. 11ost of interviewed people were directly interviewed at the company location (of-fice, meeting rooms). A smaller number of interviewees also visited us at the universityfor an interview.In order to prepare for the interview and to structure the questions to ask we createda questionnaire that contained several questions, focusing on the following main areas: • General Questions about the company: this section of the questionnaire containsthe sector in which the company is acting, the size, age of the company etc. • The Interviewee (person to be interviewed): Here, we asked the interviewedperson to give some public information about his/her position within the compa-ny. • IT Questions : Since Grid computing is not yet a term that all opinion leadersare fully aware of, we first wanted to ask some general IT questions such aswhere IT is used? Does the company use outsourcing, networking etc.? Are therestorage and/or computing intensive demands? These questions help focus andconcentrate on features that Grid computing mainly provides. • Grid Questions : The main part of the survey then focused on Grid questions,common understanding and the application of Grid technology within the com-pany or in relation to customers and their products. We also analysed if a serviceoriented software architecture is used.Typically, one or two representatives of the BIG project were leading the interviews.Each interview lasted between 30 and 60 minutes. The list of interviewed companies(which were located in Austria, Germany, Slovakia, France, and USA) is the following:Microsoft Austria, SAP Germany, Oracle Austria, Sun Microsystems Austria, BullAustria, Apple Computer Europe, Telekom Austria, Magna Steyr, WIFI Wien, MCNC,BMVIT, Uni Wien, BeoC, EC3, 1H3G (Drei Austria), IP Center, Muehlehner & Tavol-to GmbH, Uniqa, Tecco, TNT, Voest Alpine, BASF, Allianz, Austrian Social InsuranceAuthority for Business (Sozialversicherungsanstalt, SVA), The Austrian Federal Eco-nomic Chamber (Wirtschaftskammer Oesterreich)The detailed results of the interviews are confidential and elaborated in the fol-lowing section. However, after the interview series several common statements and observations were made that we can summarise as follows: • A good cost model is required. The adoption of any new (computing) technolo-gy costs money since old systems have to be replaced by new ones. In addition,Grid technology typically promises that more storage and computing power isavailable to process faster or allow for higher data availability. However, compa-nies need a simple cost model which can be used to express what is the real costcut by usage of Grid technologies. Simply put, how much money can a companysave in which amount of time? • A good accounting/billing is needed: Once Grid technology is adopted, compa-nies want to make money with the usage of Grid resources. Similar to the costmodel, a clear and precise accounting system is need that records the usage ofcustomers and charges them accordingly.12 One of the main ideas of Grid computing is the sharing of resources . However,in a business environment sharing of data creates conflicts such as “who ownsthat data?”. Some people show no willingness of sharing data or even giving itaway to be stored in remote places outside the administrative domain of a certaincompany branch. • Grid computing is often still considered to “just” provide either computing powerand/or storage. Several companies such as Sun, Amazon or Bull provide pu-re computing and storage on demand. However, customers need services ratherthan just pure CPU cycles or storage. Therefore, availability of high-level, ap-plication oriented services is very important in the commercial field. • Service Level Agreement is required for all the services offered by the Grid.This basically means that end users get a guarantee about reliable request execu-tion within some given time. Furthermore, results that are obtained by applicati-ons being executed in the Grid need to be verified. • Network bandwidth . Although the effective bandwidth of wide area networklinks is constantly increasing, the costs for commercial usage are still conside-red rather high by certain companies. Therefore, several business customers willbenefit more from Grid computing once network prices will further decrease. • Technology adoption takes time : Grid computing is currently very much rela-ted to Web services and a service-oriented architecture. Many SMEs still havenot even adopted .NET or similar Web services technologies such as SOAP etc.Therefore, several companies might remain conservative and will not adopt Webservices in the near future. • ASP profitability : Until now many CEOs in companies believe that selling soft-ware is more profitable than selling services. The business case of earning moneyby application service provisioning is (at least in Europe) not that accepted as anopportunity. However, most of the interview partners acknowledged that ASPwill be a big issue in the near future (for SMEs as well as for industry). • The term “Grid computing” seems still not to be well defined, as people ha-ve different opinions on what Grid computing exactly means. Most commonly,Grid computing is associated with “Distributed computing” and “Virtualizationof IT resources”. Many also refer to Grid computing as an “infrastructure that al-lows an efficient sharing of data across different locations”. Only some of themquoted that Grid computing is a confusing term with different meanings. Onlyone person answered that the Grid is an “infrastructure that opens new ways ofdoing business”, from which we can deduce that - as assumed - the Grid is stillassociated with technical issues rather than economical ones. Also, the visionscompanies have about the Grid, are mainly application-focused. The Grid is seenas a means to better achieve goals in specific disciplines, e.g. in biotechnology. • For the companies that completed the survey, Grid computing is an importantissue or even the main business area. It seems that those companies interested inGrid technologies put a lot of time and effort in it - and do not just run a Gridin order to experiment with it. Most of the companies are doing “Business in theGrid” by selling Grid software. They are thus on the “EG2” of the Economy-Grid Layer Model [59]. But some quoted that they were doing Business in theGrid by “using a Grid infrastructure to provide resources and/or services”, whichcould place them on the 3rd layer of the Economy-Grid Layer Model. • Conforming to the companies and research institutions, Grid technologies willcontinue to establish themselves on the market and even become more import-ant. They are said to become more important on the technical level, and 85% ofthe interviewees think that Grid technologies could have an impact on businessmodels. We can thus deduce that there is an industry awareness concerning Gridand business. The results seem to underline the estimations found in recent pa-pers related to Grid computing: until now the technical challenges prevail, butthere is awareness for the possible economical applications of Grid computing. • Grid solutions seem to be applied to many different business areas (such as Cu-stomer Management, Product Development, Supply Chain Management) andacross all industries. About 70% of those who have completed the survey provi-de IT resources or services on demand over the Grid infrastructure, and there arealready Grid solutions supporting economic principles, such as accounting func-tionalities. This result is rather surprising, as we thought that the implementationof economic principles in Grid solutions is not yet widespread.14
Finally, all the interviewed companies support research on Grid technologies -either in-house, or by providing funding, or both. It is also worth mentioning thatmany companies claim to have academic partners that do research on Grid tech-nologies, and to be in touch with international Grid organizations or communitiessuch as the Global Grid Forum.
Another outcome of the project is a Web portal for general information about businessGrids. It can be used for retrieving information collected in the project in general, espe-cially for the parts ”Revisiting of Existing Business Models” and ”Market Analysis”.The portal is based on CEWebS (Cooperative Environment Web-Services) [36].This environment provides: • The so-called Scientiki, which is a WIKI that has the additional capabilities ofexporting to L A TEX and HTML, and thus supports the creation of scientific papers. • An evaluation part that handles forms, questionnaires and surveys. • The possibility to run a discussion forum. CEWebS also provides other functio-nalities, but they were not used in this work.The portal is available at the address and contains the following four parts: • Information & Surveys : The start page of the portal contains a short projectdescription and links to the online surveys mentioned in the Market Analysisin section 5. The start page and the surveys are available both in English andGerman. As already mentioned, the surveys were presented in cooperation withthe CAS Community [35] who helped promote the surveys by announcing themin their newsletter and on their Web site. Figure 4 shows the start page of theWeb portal. • Dictionary : This dictionary is the result of a master thesis “Dictionary on Busi-ness in the Grid” [56] realised within the project . In the Web portal, all the itemscontained in the dictionary are listed and can be clicked. They are partitionedinto the following categories: – Introduction: contains general information and definitions about the Gridand Business in the Grid. – Projects: here several projects dealing with Business in the Grid are descri-bed. – Conferences, Workshops, Alliances: this section contains all the confe-rences, workshops and alliances that are considered relevant for Businessin the Grid. 15bbildung 4: Web Portal16
Companies: description of several companies providing Grid or Grid-relatedsolutions and services. – Products: here Grid products that contain some business aspects are pre-sented. – Concepts: this section contains the core of the dictionary. As in the othercategories, the items are listed alphabetically. – Links: here general interesting links to Grid resources can be found. – Conclusion and Outlook • Forum : As an additional functionality, a discussion forum is available. It ismeant for communication both among the project members and with externalpeople interested in the project. The forum supports RSS integration - users canthus subscribe to the forum with a mail program and receive the messages ase-mails. • Community : In this part the project members are listed with their e-mail addressso that interested people can easily contact them.
In the following chapter we illustrate one key business model that was identified andelaborated within the BIG project which builds an enabling basis for implementing thevision of selling software and resources as a service and not as a good. It is based on“Trust and Security” (Section 6.1) and also resulted in a prototype based on state-of-the-art Grid technology.
Trust and security are often claimed in Grid computing as some of the functional dif-ferences to earlier developments in the Web and distributed computing [20]. Beyondorganizational boundaries, virtual organizations need a trustable and secure infrastruc-ture to utilize autonomic resources and services. Security describes a field of activitiesto guarantee the privacy, integrity and availability of resources.
Although Grid research has gone a long way from the first steps in the 1980s, it isstill far from providing plug and play tools that can easily be deployed and maintained.Current Grid Solutions are mostly custom fitted systems deployed by big players likeIBM for big customers or installed at scientific research facilities fitted to the needs ofscientists that are often very different from the requirements a business poses on a newtechnology.Furthermore, most Grid projects are still very introspective. They are intended toshare the resources of some high performance computing center or provide a common17le system to members of a research group. They focus on strongly coupled rathersmall virtual organizations while the vast majority of the benefits Grid computing couldleverage keep lying bare in the Internet.Publicly available Grid services however will serve a heterogeneous group of cu-stomers not bound together by being part of the same business or working at the samescientific working group. This demands for tools to dynamically manage big com-munities of loosely coupled entities providing and consuming Grid services withoutexcessive administrative overhead.Another problem hindering the progress in publicly available Grid services is themethod of payment. Conventional services like delivering pizza or books or bookinga flight that are offered over the Web require human interaction anyway. Thereforepayment can be handled interactively by the requester of the service. However Gridservices typically only involve machine to machine interaction and should be processedtransparently for the user. Today there are very few projects that address this problem.One of the most promising usages of Grid computing is distributed storage andmanipulation of data. For a flexible and universal usage of Grid resources, it will benecessary to abstract the process of distribution and make it transparent for the Griduser.A metaphor often used for this abstraction is the electric power Grid. What seemsto the user as a simple push on a button is backed by a sophisticated framework ofpower plants and transformer stations that span (sometimes) the whole continent.To obtain this abstraction it will be crucial to have an opportunity to provide aflexible way to manage the access to Grid resources anonymously as well as trusted.While the provider of a resource wants to secure that the user of his facilities istrustable, the user himself is interested in maintaining his anonymity, while gainingeasy access to the services.Staying with the metaphor of the electric power Grid, the user does not want tohave to identify himself to an operator of the electricity provider if he turns on the light(and therefore uses their services). In turn the electricity corporation does not want toface the bureaucracy arising with identification for every watt they send him. Insteadthey are interested in knowing him as a reliable customer with a sufficiently coveredbank account to pay for his demand of electricity.In the course of the BIG project we developed gSET [58], a mapping of SecureElectronic Transaction (SET) [37, 38, 39] to the Grid environment, which can fulfillthese requirements.It is intended to show a possibility to handle authorization and payment in looselycoupled Grids of commercially interacting units. Using Grid services should be as easyas paying with a credit card (or even easier).The method described provides an authorization mechanism that is request based and dynamic . Request based means that the authorization to use a service is evaluated at thetime the service is to be invoked and can depend also on the contents of the serviceinvocation not only on the role and rights of the service requester.Furthermore in gSET the authorization to use a service depends on the service re-quester’s credit at a configurable account provider, making the solution highly dynamic and easy to integrate into existing accounting systems. Due to this delegation a gSET18nabled service does not require an extensive account management by the service pro-vider and is therefore perfectly suitable for the commercial service provision. gSET aims at providing a framework that enables a provider of a Grid service to makehis service accessible for everyone willing to pay the price set by him.gSET must address the following important issues: • Relieve the service provider of complex account management and payment issu-es • Secure the transaction against observing and manipulating attacks • Protect the service requester’s confidential data against fraudulent usageTo relieve the service provider of account management and payment issues whileguaranteeing the trustability of the service requesters, we have to delegate these topicsto a third party. Like in current credit card payment solutions, we need a paymentservice that evaluates the credit of the service requester and manages the transfer ofcredits.To secure the transaction against observing and manipulating attacks we must usea combination of digital signatures and encryption. Fortunately Grid environments al-ways contain public key infrastructure so we can assume each participant owns anidentity certificate along with a public/private key pair.The protection of the service requester’s confidential data against fraudulent usagehas two aspects. Firstly we do not want the service provider to know payment relatedinformation like the account number or a possible password. Secondly the paymentservice should not receive information about the nature and the parameters of the con-sumed service.We can now identify three actors in our scenario, the service provider , the servicerequester and the payment service . If we look at current electronic payment solutionshowever, we find that the payment service is in most cases split up into two closelyrelated but nevertheless independent units. In our case we will call them account pro-vider and trust manager . While the account provider maintains the account and therelationship to its account owners, the trust manager is responsible for the technicalexecution of the payment process. The architecture of the gSET actors is depicted byFigure 5.If we take a closer look at the situation we can identify five use cases. • The Service Provider contracts a Trust Manager to handle the authorization andbilling. • The Service Requester contracts an Account Provider to vouch for him. • The Service Requester requests the price for the Service. • The Service Requester uses the Service.19bbildung 5: gSET architecture • The Service Provider collects his credits.With gSET the latter three use cases are addressed, while we assume that the firsttwo are already successfully provided by the Grid infrastructure.Before the service requester uses the service he must have the possibility to retrievethe price the service provider will charge him for the requested operation. To get thisinformation the service requester must send information on how he wants to use theservice.When the service requester decides to use the service, he must first request autho-rization. To obtain authorization he must send the information on how he wants to usethe service along with which account provider he wants to use, his account details, andthe amount the trust manager should be allowed to authorize. It must be guaranteed,however, that the account details and the amount to be authorized are not visible to theservice provider, nor can be manipulated by him.The service provider relays the account details and the amount to authorize (but notthe information on how the service requester wants to use the service) to a trust managerthat can manage transactions for the service requester’s account provider along withthe amount he intends to charge the service requester. The trust manager checks if thecharged amount lies within the limit given by the service requester and the credit theservice requester is granted by the account provider. If both checks succeed he storesthe information on the amount the service provider charged and sends a positive replyto the service provider along with some kind of token that the service provider can useto collect his credits later. If the service provider receives a positive reply he allows theservice requester to use the service in the desired way.
Let us visualize this scenario: Jim is an ambitious tourist who wants to record hismemories by taking photographs of all the historical places of Vienna by using his3.2 Mega Pixel camera built into his cellular phone. But soon after a few pictures, he20ealizes that he has run out of his mobile phone memory. He suddenly recalls the Gridservices accessible by a broker. He seeks the address of the broker from his bookmarksin his mobile and sends it a request for storage of pictures that he has already taken. Thebroker responds promptly and sends him links of different Service Providers and theirrates. When Jim selects a Service Provider, the Service Provider responds back and thepayment process uses Jims credit card. Jims credit card information is already savedin his mobile device but he is confident that this is a very safe process and there is noneed to worry about anything. He realizes that the data has started migrating, but whata pleasant surprise! He notices that the mobile device is using Bluetooth to move thedata, thus no fee will be charged for the transmission. After the successful migration hereceives some thumbnails of the images which are in fact tickets that he can use lateron to retrieve his pictures. He wants to send some pictures, that he has already taken, tohis friends. He instead sends some tickets of the pictures through SMS. After a coupleof days he uses the rest of the tickets to retrieve his data.In the above architecture we make it possible to sell and buy Grid services throughcell phones. Our business model is based on simple and straight forward principles. • The customer can buy things without revealing her/his financial information tothe vendor. • Customer and vendor both express their trust in a mediator who assures the cu-stomer about the reliability of the vendor and guarantees the vendor about thefinancial status of the customer.We propose a novel business model through our state of the art financial mechanismcalled gSET . gSET keeps the financial information separate form the order informa-tion and enables both parts of the information to reach the appropriate players i.e. theTrust Manager and the Service Provider respectively. The Trust Manager plays a veryimportant role in a financial transaction in the sense that both the Service Provider andthe Account Provider have trust in it. The Account Provider verifies the client’s finan-cial status and thus the Service Provider has no hesitation to process the order. In thescenario that we have presented here, the customer buys storage services from a ven-dor. Customer’s financial credibility is maintained with the Account Provider that maybe at the same time the mobile service provider and charges the customer on a regularbasis. The Account Provider pays for the service on behalf of the customer and thisamount is added to the monthly bill of the customer. Account providers need to have acontract with the Trust Manager. The broker can offer cheap and premium accounts tothe Service Providers and charge them on a regular basis. Customers can rank ServiceProviders according to service reliability. This ranking can also be done based on salespatterns. A more reliable Service Provider gets a higher position on the list of ServiceProviders that is sent to the customer by the broker.The gSET model highlights five dimensions of doing business on the Grid. • Transparency.
The gSET model enhances the concept of transparency in theGrid. The service provider and the user are unaware of each other while theTrust Manager plays a mediating role and helps to establish trust between thesestrangers. There is a fool proof payment mechanism that fits seamlessly in theservice oriented architecture. 21n order to achieve this kind of security, gSET could be of great help. Whilethe encryption of the remotely stored data is not defined in gSET, it can helpproviding means of distributing sensitive data under relative anonymity whileholding the key to putting it all together again. This minimizes the data to bephysically protected and helps enhancing security. • Trust.
The gSET model provides the bridge of trust between two completelyunknown business partners i.e., the user and the service provider. This trust isguaranteed by a very secure PKI based communication mechanism having thetrust manager as the focus. Both parties express trust in the trust manager whichguarantees the credibility of both parties and yet keeps their confidential infor-mation secret. • Privacy.
The service provider does not know the credit information of the userand at the same time the account provider is also unaware of the shopping-historyof the user. In our example we have shown how to use the PKI infrastructure ofthe gSET to encrypt the information stored on the service provider’s devices.With gSET, the service provider obtains only as much information about theservice requester’s identity as the service requester intends to give. However,this relative anonymity does not deprive the service provider of the trust in theservice requester’s reliability, as the trust is delegated to the trust manager.The trust manager does not receive information about the details of the servicesthe service requester uses. If the service requester wants to put the level of pri-vacy protection even higher he may even use more than one account provider tovouch for him. • Agility.
The gSET model is a very secure, sound and yet flexible system as it canbe fit within an already designed workflow. It acts as a loosely coupled paymentmodule that takes care of financial transactions among business partners on theGrid. We have demonstrated [49] how to fit this module into already definedbusiness workflow on the Grid.With a natively distributed solution like gSET, it is easier to address the challen-ges of a mobile world. • Reliability.
The gSET model is in fact a gridified version of the SET model[37, 38, 39]. SET was a very reliable, errorfree financial transaction system withdisconnection and rollback facilities. SET enhances security and reliability by itscertificate distribution. This distribution of certificate although being a very po-werful mechanism was not accepted as it was realized cumbersome by the usersand hence SET couldn’t succeed. The case with Grid is different as it alreadyimplements security by distribution of certificates among the partners. ThereforegSET sounds like an ideal match for the GridThe benefits of using gSET as authorization mechanism in business scale storageservices are evident for both service provider and service requester.The service provider does not have to meddle with a complicated account manage-ment. He can easily delegate the issue of the service requester’s trustability to a trust22anager. This allows the access to a much bigger market, as the account providers canattract a lot of customers willing to express trust in trust managers because they do notwish their financial information to be disclosed all the times.The only requirement regarding certificates, gSET poses, is that both partners trustthe certificate of the trust manager. In an economic market of service providers, thepolicy is quite simple. The provider sells to anybody who does not have any legal orrating problems. For this, the provider needs to ensure that the customer is trustable andpays for the used service. With gSET it is possible to transfer the permission to verifythis information for one specific transaction without disclosing the private information.
During more than two years lifetime of the BIG project, Grid standards and technolo-gies have improved quickly, and several new business models have been investigatedby the BIG team. Although Grid technologies are well established in certain commu-nities in research and academia (data and computing intensive science fields such asphysics, bioinformatics, etc.), engineering etc., the commercial market is not yet readyto adopt Grid technologies. This has several reasons such as evolving standards andpartly immaturity of Grid solutions. We see a trend that the Grid is still in the phasewhere early adopters and expert users are the dominant users. However big softwarecompanies are promoting Grid technologies which is a good sign. Finally we presenteda specific business model for the mobile Grid, which is based on gSET allowing fordynamic authorization. gSET is the first enabling step to make Grids a platform forcommercial workflows.
Acknowledgement
This research project was funded by project number 10547 of the OeNB AnniversaryFund.
Literatur , page 245a, Nice,France, April 22-26, 2003. IEEE Computer Society Press.[12] William H. Bell, David G. Cameron, Luigi Capozza, A. Paul Millar, Kurt Stockin-ger, and Floriano Zini. OptorSim - A Grid Simulator for Studying Dynamic DataReplication Strategies.
International Journal of High Performance ComputingApplications
Economic-based Distributed Resource Management and Sche-duling for Grid Computing TM IFIPInternational Conference on Network and Parllel Computing , 2005.[22] Ian Foster and Carl Kesselman, editors.
The Grid: Blueprint for a New ComputingInfrastructure . Morgan Kaufmann, San Francisco, California, 1999.[23] Christopher Freeman.
The economics of innovation
SET Secure Electronic Transaction Specification, Book 1:Business Description , May 31, 1997.[38] MasterCard, VISA.
SET Secure Electronic Transaction Specification, Book 2:Programmer’s Guide , May 31, 1997.[39] MasterCard, VISA.
SET Secure Electronic Transaction Specification, Book 3:Formal Protocol Definition
Oracle Grid Index Report , 2nd edition, April 2005.2546] Thomas Sandholm, Peter Gardfjll, Erik Elmroth, Lennart Johnsson, and OlleMulmo. An OGSA-Based Accounting System for Allocation Enforcement acrossHPC Centers. In ,New York, USA, November 15-19, 2004.[47] Erich Schikuta. NeuroWeb: an Internet-based neural network simulator. In , pages 407–412, Washington D.C., November 2002. IEEE.[48] Erich Schikuta, Flavia Donno, Heinz Stockinger, Elisabeth Vinek, Helmut Wan-ek, and Thomas Weishaeupl. Business In the Grid: Preliminary Project Results.In , OCG Verlag, Hagenberg, Austria, Dec. 1 - 2,2005.[49] Erich Schikuta, Jrgen Mangler, Irfan Ul Haq, Helmut Wanek, and Oliver Jorns. Agridified, secure, mobile business workflow using gset. In
Workshop on EconomicModels and Algorithms for Grid Systems in conjunction with the 8th IEEE/ACMInternational Conference on Grid Computing (Grid 2007) , Austin Texas, Sep-tember 2007.[50] Heinz Stockinger. Defining the grid: A snapshot on the current view.
Journal ofSupercomputing
Business in the Grid: BIG Project , 2004-2005.[56] Elisabeth Vinek. Dictionary on Business in the Grid. Master’s thesis, Faculty ofComputer Science, University of Vienna, January 2006.[57] Thomas Weishaeupl, Flavia Donno, Erich Schikuta, Heinz Stockinger, and Hel-mut Wanek. Business In the Grid: The BIG Project. In
Grid Economics & Busi-ness Models (GECON 2005) of Global Grid Forum 13 (GGF13) , Seoul, Korea,March 2005.[58] Thomas Weishaeupl, Christoph Witzany, and Erich Schikuta. gSET: Trust Mana-gement and Secure Accounting for Business in the Grid. In
Sixth IEEE Interna-tional Symposium on Cluster Computing and the Grid (CCGrid 2006) , Singapore,May 16-19, 2006.[59] Thomas Weishupl and Erich Schikuta. Towards the Merger of Grid and Economy.In Hai Jin, Yi Pan, and Nong Xiao, editors,
GCC Workshops , volume 3252 of