Service provision and Quality of Service metrics: the TINA enterprise viewpoint Zoran Milosevic Computer Science Department & Cooperative Research Centre for Distributed Systems Technology, The University of Queensland, QLD 4072, Australia Tel: +61 7 365 1368, Fax: +61 7 365 1999 e-mail:
[email protected]
Abstract The focus of this paper is on the TINA enterprise viewpoint. We propose an extension of this viewpoint with relevant concepts taken from current Open Distributed Processing (ODP) standards. This is illustrated by analysing the telecommunication service provision process using the TINA enterprise specification. Additionally, this viewpoint is utilised as a starting point for addressing the complexity of new Quality of Service (QoS) issues in TINA consistent platforms. This is augmented by an emerging awareness that new QoS requirements encompass both technical and economical aspects. In order to provide appropriate quantitative measures for QoS levels, a general, application-independent QoS metric framework has been proposed. This framework is derived from an economic theory of consumer demand. It can be used by any party involved in the service provision process, e.g by end-users to measure their satisfaction and hence, be able to compare different service providers, and by service providers to determine their competitiveness.
1
Introduction
In order to facilitate the merging of telecommunications and computing into an overall information networking architecture, the TINA initiative capitalizes on new and promising technologies from both fields, i.e. broadband transmission/switching and distributed processing. Broadband technology provides a basis for new media types and many new services which can be brought to Customer Premises Equipment (CPE) in a flexible and cost-effective manner. Distributed processing offers mechanisms to efficiently share the resources in such a network in order to improve many of its characteristics including, performance, reliability, availability, flexibility etc., and to increase sharing of common services. With an objective being to promote interoperability between heterogeneous technologies and different vendor products, the TINA initiative and the recently established TINA consortium aim to build on and maintain compatibility with relevant standards in distributed processing and telecommunications[1]. Examples of such standards are Intelligent Network (IN), OSI Network Management, Telecommunication Management Network (TMN) and Open Distributed Processing (ODP). It is also expected that international standards will benefit from the results/ experience of the TINA initiative, which is another TINA objective. However, in addition to these technological trends, information networks of the mid ‘90s and beyond must also take into account another factor, i.e. market forces. This is an important issue, especially in emerging open environments (where open means more than purely technical issues, i.e. open access to the public). Indeed, this factor reflects customers’ demand for information services and thus dictates business objectives of information service providers. Its importance has emerged from the deregulation of telecommunications and the commercial push towards open systems in computing.
The environment in which future telecommunication/information services are to be provided/executed is likely to utilize a common underlying software infrastructure, such as that advocated by the TINA Platform Reference Model (TINA-RM) [2]. In addition, the service provision process will involve various parties with different business objectives. This combination of complex technological and economic forces motivates us to apply the TINA viewpoints approach to analyse the telecommunication service provision process. These viewpoints are enterprise, information, computational and engineering. This approach can be used to master the complexity of information networks (the ‘separation of concerns’ principle) and is similar to the one used within the ODP context[3], but with more telecommunication specific requirements. Most work on TINA so far has been focused on information, computational and engineering issues. This is understandable, as they have more (traditional) ‘architectural’ relevance and thus, have attracted more technical-related interests of telecommunication operating companies and computer/telecommunication manufacturers. However, the enterprise viewpoint issues must not be neglected as they represent the starting point in addressing enterprises’ critical business objectives, as determined by market forces. In order to highlight this, we apply the enterprise viewpoint to a specific enterprise objective, i.e. the Quality of Service (QoS)1, in its entirety (including technical and economic issues). Similar to the Reference Model of ODP (RM-ODP) [3], the TINA enterprise viewpoint should serve as the basis for specifying enterprise goals (and associated policy statements) on which all other viewpoint specifications will directly or indirectly depend. For example, in order to meet an enterprise QoS objective such as response time, the information model may require a different partitioning of its information, the computational model needs appropriate computational algorithms, and the engineering model may require replication of data. The focus of this paper is on the TINA enterprise viewpoint. To illustrate some important enterprise issues, in section 2 we deal specifically with an analysis of a typical telecommunication service provision process and identify the points where an important enterprise issue, i.e. QoS arises. In section 3 we outline new, TINA-related QoS requirements and propose a general QoS metrics framework which can be used when dealing with the complexity of QoS issues. In section 4 we apply this framework to two service examples, namely a tourist information service and the TINA contract trading service. Conclusions and future work are outlined in section 5.
2
Telecommunication service provision - enterprise viewpoint
The TINA enterprise viewpoint describes the overall objective of a TINA consistent network within an enterprise (possibly including several cooperating organisations) and utilizes concepts such as roles, actions, goals and policies. We propose an extension of this description with current concepts from the ODP standards [3]. These include enterprise objectives and policy statements (for ODP systems), as well as roles, activities and obligations undertaken (by objects in the ODP system)2. We apply these concepts to the 1. For the definition and other details related to QoS, see section 3. 2. Note ODP enterprise related terms are in bold type in this section.
analysis of the telecommunication service provision process and illustrate how they can contribute to consistently addressing QoS issues in a TINA environment. Telecommunication service provision is the enterprise objective of a community. This community may involve various agents (agent is ‘an object which has a role of involving performative actions’ [3]). The objective is a contract of obligations between roles fulfilled by agents. The following agents, which take part in the service provision process can be identified (Fig. 1), with their associated roles: • Network provider (carrier) - an agent which offers network resources (i.e. objects whose existence is subject to accounting[3]), such as transmission facilities (e.g fibre trunks, mobile and satellite systems), switches (e.g. digital switches for voice/data, often with support for IN type of services, such as the Service Switching Point, SSP), control nodes (e.g the Service Control Point, SCP in INs), management facilities (e.g. Operations Support Systems, OSSs) and relevant services. Additionally, carriers frequently offer Value-Added-Services (VAS), such as the IN-type of services. The carrier is the owner of these resources/services whose use is the subject of the carrier’s charging policies. • Service provider - an agent which offers services on top of the network and which may or may not be an owner of network resources. The service provider normally purchases network resources and/or services from a carrier and uses them to offer its own VASs, for which its charging policies apply3. • End-user - an agent which directly experiences the effects of services, e.g. one which uses services such as toll-free, unique national number etc. This is a fairly general service provision scenario. However, it is sometimes important to identify another type of the service provider’s potential customer, i.e.: • Service subscriber - an agent which is a purchaser of a service provider’s service and which makes these services available to end-users. For example, agents which offer to their customers services such as toll-free, unique national number services, are purchasers of these services from service providers. Each agent has a set of enterprise objectives, subject to constraints/policies set by national/ international regulatory authorities, i.e. policy makers. Possible enterprise objectives can be the maximization of variables such as profit, profit rate, enterprise growth opportunities, QoS value, performance/cost ratio and an increase in enterprise prestige[4]. Being an objective of the community, service provision should be seen as governed by a set of policies (policy in enterprise specification expresses how communities reach their objectives [3]). An important set of policy statements which are addressed in this paper relate to enterprise specific QoS issues, regarded as an important business link between agents. Interactions between agents are determined by two types of performative actions, i.e. one agent provides a service to another, for which the latter pays a certain amount of money. 3. Service providers may simply resell network provider’s services.
End-user pay pay
QoSSs,Eu
pay
Service subscriber pay
pay
QoSSp,Ss
QoSSp,Eu
Information service provider pay
QoSNp,Ss
QoSNp,Sp
QoSNp,Eu
Network provider
Figure 1 Therefore, one agent is a service ‘producer’, while the other is a service ‘consumer’. An agent can have both consumer and producer roles, e.g. a service provider normally has a consumer role with respect to a carrier, and a producer role with respect to end-users/ service subscribers. Basically, a consumer of a service is concerned with the value obtained from this service (which directly depends on the level of QoS provided), and a service producer, with the (monetary) value received from the consumer. Thus, in the service provision process, this QoS/price pair binds agents which take part in this process. The practical implementation of this binding process is a policy statement in terms of a contract formed between agents after QoS negotiations are carried out. This contract covers issues such as guaranteed levels of QoS, corresponding price and responsibilities of agents. As can be seen from Fig. 1, there are several points in this process where pairs (QoS, price) determine relationships between agents. These are annotated as QoSSp,Ss, QoSSp,Eu, QoSSs,Eu, QoSNp,Sp, QoSNp,Eu, and QoSNp,Ss. Since different agents have different roles (and objectives) in service provision, there are also different agent-specific QoS requirements which have to be met in order to make a consumer agent satisfied. For example, the
service provider has to assure that the service delivered to the service subscriber meets an agreed upon QoS level (QoSSp,Ss) and price4. The QoS performance parameters in this case depend on the type of service delivered (in section 4 we elaborate with an example of a specific service). However, the service provider has to guarantee a satisfactory level of service to end-users (QoSSp,Eu), as their service usage determines the revenue of service subscribers, whose satisfaction in turn, determines the service provider’s revenue. Similarly, the network provider has to deliver appropriate levels of QoS to its consumers, e.g. service providers (QoSNp,Sp) and end-users (QoSNp,Eu). Namely, the network provider has to make sure that service providers are satisfied with the network services delivered, but also, since the network is normally an access point for end-users (and possibly service subscribers), the network provider needs to satisfy all parties using the network[5]. The agents mentioned above can cooperate in order to achieve their individual objectives. We assume that an underlying TINA-RM provides a supporting environment in which this can be achieved. The central TINA-RM enterprise concept which provides this is: • the contract trader5, i.e. an agent with a role to accept offers from service producers (e.g service providers) and to respond to a service request from service consumers (e.g end-users and/or service subscribers), so that an appropriate (and possibly the best) service producer can be found. Basically, the trader’s role is to serve as an advertising domain for service producers; this advertising facility is consulted by service consumers when pursuing service of a particular type and with specific attributes. As such, the trader is a central component which supports open information markets. Since the trader service itself may be owned by different agents[6], there can be various scenarios with respect to the position of this service in the service provision process. For example, if the trader is owned by the network provider it can be used by service providers to advertise their services and by endusers to look for appropriate service offers (a simplified scenario for this case is depicted in Fig.2). If, on the other hand, the trader is owned by a specialised trader service provider, it can also serve as a facility for network providers to advertise their resources/services. Similar to other agents, the trader has its own QoS specification and associated charging policies for using its resources. For example, end-users/ service subscribers will be satisfied if the trader provides them with a satisfactory QoS level, such as matching parameters that accurately reflect desired service attributes, flexibility in terms of choice, satisfactory response time etc. On the other side, service providers will be happier if customer’s usage of trader is high, and if greater coverage is provided etc[6]. Applying the TINA enterprise viewpoint in this manner provides a clear identification (and separation) of various concerns of parties involved in associated service provision/ execution, and may serve as a good starting point for further analysis of an aspect of this process. Specific aspects of this paper which are also within the scope of the enterprise viewpoint are QoS metrics and related charging issues.
4. Normally service subscriber is a direct customer of a service provider [5]. 5. Similar to the ODP trader component. The term contract comes from the TINA concept called contract (see the Appendix).
Contract trader Trading
Charging
QoST,Eu QoST,Sp pay pay
End-user
pay QoSSp,Eu
Information service provider
Figure 2
3
QoS metrics issues - enterprise viewpoint
Technological and economical factors relating to new information environments (such as the environment advocated by the TINA initiative), are driving forces for the changing view on QoS issues, previously adopted in traditional telecommunications. A consensus is gradually being reached where the QoS framework needs to be more general so as to include new factors such as, support for different types of consumers, variety of service classes, customer-driven (not network provider supplied) QoS requirements etc. 3.1
New TINA-related QoS requirements
We highlight several new QoS requirements which a TINA consistent platform will bring. First, QoS issues should be end-user driven, with a meaningful definition of terms for both end-users (service consumers in general) and service producers. Second, QoS should be analysed on an end-to-end, application specific basis, whereas the traditional communication QoS (in terms of delay, throughput etc.) is just one element of this. For example, additional TINA platform related details, which are of particular TINA viewpoint concern, should be included in the TINA QoS picture. Further, QoS negotiation, QoS guarantees and dynamic QoS aspects need to be taken into account. Thus, there is a need for a broader scope of QoS issues which should include both technical and economic factors. The economic factors emanate from an individual agent’s enterprise objectives, which often depend on QoS levels. For service consumers QoS determines satisfaction and for service producers, it determines competitiveness. However, provision of a service of high quality implies a high price to be paid. The (QoS, price) combination plays a major (business) role in binding agents together. For example, users are interested in maximizing their perceived QoS/price ratios, and service providers in maximizing their offered QoS/cost ratios.
3.2
QoS metrics framework proposed
In order to meet these new requirements, a QoS metric is needed to provide quantitative measures for different levels of service quality, i.e. different service performance parameters, as well as a measure for a total QoS, based on these individual parameters. Such a metric is of particular importance for consumers (e.g end-users), as they pay for the QoS delivered and thus need a means for comparison between different service providers. It can also assist service producers in identifying and appropriately quantifying particular service parameters relevant for consumers, as this may improve customer satisfaction and consequently enhance the provider’s competitiveness. Finally, regulatory bodies may also need this metric when prescribing statutory requirements for QoS monitoring and reporting[7]; they can assess the effects of deregulation (in terms of the maximization of benefits to consumers) and be able to promote efficient operation of markets by improving the information base available to consumers [8]. In order to provide a comprehensive generic QoS metric framework we develop a methodology which is based on an economic theory, i.e Lancaster’s theory of consumer demand[9]. This theory is based on two fundamental propositions, whereby it is possible to clearly distinguish between the relationship of goods and their characteristics and the relationship of characteristics and people, i.e. objective and subjective relationships. The first proposition states that all goods possess objective characteristics relevant to the choices people make between different collections of goods. The second proposition states that individuals differ in their subjective reactions to different characteristics, i.e. it is the characteristics of goods, not the goods themselves which people are interested in [9]. Therefore, goods and services can be viewed as bundles of characteristics and quality refers to variations in quantities of characteristics[8]. For the purpose of practical studies (i.e. to produce simple measures with minimal cost) Lancaster suggests to first identify major areas of consumers’ concerns, i.e. service aspects[9]. Service aspects are application specific (e.g. in the case of a directory service these are, searching, information presentation, billing etc.). Second, each of the aspects’ relevant characteristics should then be identified. For example, service characteristics may be, flexibility, simplicity, accuracy, availability, reliability, responsiveness and security6. QoS aspects and QoS characteristics, can then be represented as rows and columns of the so called QoS matrix. Third, each row/column pair (i.e. matrix cell) gives a relevant performance parameter (service indicator), which is an objectively and/or subjectively measurable variable. Objective variables are directly measurable parameters, e.g. number of failures, end-to-end delay etc. Subjective indicators are not directly quantifiable, e.g. user friendliness, understandability of billing etc. Rather, they represent a level of consumers’ satisfaction, usually in terms of the percentage of consumers satisfied with a particular aspect, or in terms of a rating scale (say 1 to 10) of consumers’ satisfaction. Subjective indicators are also important as they can impact on the competitiveness of service providers. Once the previous steps in the empirical QoS study have been carried out, the last step entails the implementation of mechanisms for verifying accuracy of QoS statistics (provided by carriers) and for reviewing or 6. It seems that these characteristics are general enough to be applicable to any TINA application. Thus, the main task is really in identifying appropriate service aspects.
enforcing QoS standards [8]. To treat QoS using such a matrix is appealing, since the matrix can be: • used to specify relevant performance parameters for any type of service, i.e. it is service (or application) independent, though the particular elements of the matrix will vary between services. • applied at various points in the system, where the pair (QoS, price) determines different business arrangements among the agents involved • utilized by any agent concerned about QoS issues (e.g. end-user, service provider, regulator etc.) • utilized when dealing with QoS issues from any of the relevant TINA viewpoints. From the enterprise viewpoint, the QoS matrix can be seen as a unifying business link between consumers and network/service providers. Consumers first state their QoS requirements in terms of the cells’ performance parameters. These should then be included as a part of the service/network provider’s enterprise specification, which in turn can be used as a starting point for the specification of other TINA viewpoint concerns. There are several issues which need to be addressed when adopting the above approach. First, it is important to correctly identify a small group of service aspects which have a major effect on consumer utility, i.e. to find mechanisms for inferring the value placed by consumers on relevant aspects of service quality [8]. Some methodologies normally used for this are consumer surveys (ranking of relative importance of individual aspects), choice models (values that consumers implicitly use for alternatives while trading off price with various service aspects), judgment of experts [10] etc. The choice of an appropriate method for telecommunication services will depend on factors such as availability of data and the nature of the service being studied. For example, the consumer survey may be a widely acceptable method from the customers point of view, but for more complex applications an expert’s judgment would be more appropriate due to the possible technical complexity of such applications. Second, the proposed framework includes both objective and subjective types of performance measures (indicators). The objective measures directly relate to the quantifiable characteristics for a particular service/product aspect and thus measure the associated performance parameter delivered by the service producer. They also incorporate subjective elements, since the identification of key service aspects is based on consumer preferences. However, as stated earlier, the matrix also includes subjective indicators; they are required because objective indicators are not sufficient to measure all aspects of consumers satisfaction. In other words, the mapping (relation) of objective parameters to consumers’ satisfaction level is sometimes difficult to ascertain in practice. For example, objective parameters cannot be used to measure the courtesy of telephone operators or the cultural or aesthetic value of television programs[8]. Therefore, the subjective indicators are used when there is a need to directly measure consumer’s satisfaction with some service aspects. The optimal balance of subjective and objective indicators may depend on the purpose and audiences for which the indicators are prepared. For instance, in the competitive environment, service producers may emphasise subjective indicators, while regulatory bodies will be more concerned with
objective indicators[8]. Finally, we make some observations on deriving a total QoS measure (or composite index [8]), based on individual performance parameters (indicators). This measure represents a single quantification of service quality which assigns weighting coefficients (reflecting the relative importance of certain aspects) to individual indicators. This is regarded as a very attractive method, due to its simplicity and ease in identifying trends in the overall level of service quality. However, it is important to note some limitations and approximations of the method. It is not easy to obtain the precise value for the coefficients, due to the limitation of methods for estimating consumer’s priorities[8], as stated above. Additionally, these weights will vary between different consumer groups, as a result of their different priorities. Hence, a composite index can be used to provide a useful global measure of service quality, once these limitations are recognised [8].
Tourist Information Directory
Air-fares
Package tour deals
special national international
business travel
tourist
Accommodation motels
hotels
caravans
Figure 3
4
Examples of the application of the QoS framework
In order to illustrate the applicability of the proposed QoS framework to the TINA environment we choose one specific application (envisaged as available in the near future), i.e. a tourist information service, and a TINA service, i.e. the contract trader service. 4.1
QoS metric framework - a tourist information service example
We assume the following scenario. A tourist information service (TIS) will be provided in the future that will allow customers to have access from their homes, or their working organizations, to a multimedia tourist information directory. This directory has a wide range of related information, a subset of which is presently available from traditional tourist agents (an example is shown in Fig.3). Customers will normally be able to perform the following functions: • searching through directory information in order to find particular types of information.
• information presentation of particular information found after searching. We assume that this may involve a multimedia type of information related to a certain aspect. For example, a video presentation of a particular resort, hotel rooms, surrounding areas, potential entertainment guide, accompanied with high-quality audio information, music etc. • booking of specific accommodation, flight etc. • access to various forms of billing information. • customizing of the TIS service according to customers’ own needs. Following the enterprise viewpoint, major agents centred around this service can be identified, i.e.: • An end-user of the TIS service, who is directly using it and who pays for its usage. • The TIS service provider - an agent (in fact a broker) who owns and offers the service. Examples are: a large (may be global) organization, either being a private company (such as a tourist agent, air line etc.) or public authority (such as a government tourist organisation). • A network provider who provides broadband transport services, necessary for the implementation of multimedia services. • TIS service subscriber (a subcontractor of the TIS service provider) who provides common tourist services, such as air-fares, accommodation, car-rentals etc. We illustrate the applicability of the general QoS framework (as described in section 3) by selecting the TIS provider/end-user relationship. In this case, QoS aspects which are of relevance for the end-user are, searching, information presentation, booking, billing and possibly customizing operations, as mentioned earlier. The following set of user related QoS characteristics are identified: flexibility, simplicity, accuracy, availability, reliability, responsiveness and security. The QoS aspects and characteristics are grouped to form a table with the format given below (service aspects are depicted as the rows and service characteristics as the columns of the table). For each of the aspect/characteristic pairs, the appropriate performance parameter (i.e. service indicator) is identified. Some possible indicators are given in the table. It is worth noting that there may be a significant variation in importance of specific service aspects for different groups of consumers. For example, there may be differences between business and tourist travellers with respect to the value they assign to the number of options for booking, billing, etc. This is expressed by appropriately assigning weight coefficients to particular indicators.
Table 1: TIS QoS matrix Flexibility
1
Searching
Options
Information presentation
Number of media types supported
Simplicity
Accuracy
User friendliness2
-% of correct information - level of currency of information
User friendliness4
- jitter regulation - streams synchronisation - rate regulation
Availability
- coverage
3
Reliability
Responsivn.
-% of incorrectly found information - % of cutoffs
database access time
% of cutoffs
- set-up delay - transmission delay
Security
Booking
Options5
User friendliness
of travel components6
of information7
booking time
- % of leaked information to competitors
Billing
Options 8
Understandability level
Correctness9
% of optimal
accounting time
- % of fraud - level of confidentiality
Customizing
Options10
Level of adaptability
1 In terms of a number of various search constraints (e.g to find the most convenient travel sched-
ule with the cheapest price). 2 Level of user-friendliness, measured by ease of use. 3 In terms of number of TIS subscribers included 4 Measured by ease of use, conformance to standards (e.g do all media types conform to some commonly used channel types, such as Standard Video, Slow Scan Video, Voice Audio, hi-fi Audio), etc. 5Number of options offered (e.g. advanced booking) 6 Level of accuracy of the booked information (e.g. various scheduling times, prices, alternatives etc.) 7 Number (%) of inconsistent information (e.g. unreliable ‘the last available seat’ information). 8 Number of different ways of paying (e.g. credit cards, vouchers, cheques etc.), of various formats for billing reports (e.g. frequency of periodical reports, record logs etc.) 9 The precision of accounts, travel summary data accuracy etc. 10 Number of available options for a user to customize the service to her own needs (e.g. preferred way of paying, always searching the same domains, customized booking procedure etc.)
4.2
QoS metric framework - the contract trader example
Similar to the ODP-RM, TINA platform includes a trader component (i.e. the contract trader), that provides a (contract7) registration service for service producers (servers) and a contract selection service to service consumers (clients). We assume that a contract trader may also incorporate a more sophisticated functionality, which the ODP trader provides. For example, a contract trader can optionally select the best contract (service 7. For the definition of terminology used by TINA, refer to the Appendix.
offer) for a client from the set of possible matches, as well as linking up (with traders from different administrative domains) to form a federation so that a wider market for service producers and a greater choice for service consumers can be achieved. The following trading service aspects are identified: contract registration (similar to ODP exporting), contract selection (similar to ODP importing), federating, charging, and customizing to client’s/server’s requirements. For the trading service characteristics we adopt the same set as in previous example, i.e. flexibility, simplicity, accuracy, availability, reliability, responsiveness and security. These aspects/characteristics are grouped to form the QoS matrix (Table 2). Each matrix cell gives a relevant trader performance parameter. Some possible performance parameters are also given in this table. Table 2: Contract trader QoS matrix Flexibility
Simplicity
Accuracy
Availability
Reliability
Responsiv.
- coverage2
% of cutoffs
- registration /update delay
%of leaked information to other servers
coverage2
- %of correct matching -% of cutoffs
- set-up delay - information provision delay
% of access to protected information
- %of correct matching -% of cutoffs
- set-up delay - information provision delay - registration/ update delay
- % of fraud
level of timely accounting
-% of fraud - confidentiality
Contract Registration
Options1
User friendliness
- % of incorrect information - level of cur rency of information
Contract selection
Options3
User friendliness4
% of incorrect matches/ selections
-
Federating
Options5
ease of linking traders
% of incorrect information
- coverage2
Charging
Options6
Level of Understandability
level of correctness of accounts7
- of various formats8
Customizing
Options available9
Security
Adaptability
1 In terms of number of service types available for registration. 2In terms of number of potential servers or clients that can be supported 3 Number of options. They could be: provision of a selection, provision of Value Added Services,
choice of searching algorithms/selection criteria, provision of approximate matching/selection, choice of search limits for time/quantity etc. 4 Level of user-friendliness- measured by ease of use, conformance to standards etc. 5 In terms of extent (%) of trader database and service types an exporting trader is willing to share with an importing trader 6 Number of ways of paying (e.g. credit cards, vouchers, cheques etc.) 7 The precision of accounts, service usage data accuracy etc. 8 Number of formats for billing reports (e.g frequency of periodical reports, record logs etc.) 9 Number of options available for a user to customize the service to her own needs (e.g. preferred way of paying, always searching the same domains, customized searching procedure etc.)
Note that for these relatively complex applications, the matrices are fairly ‘full’. An empty cell means that the corresponding aspect/characteristic pair is not relevant for the
application. Also, some of the values in the matrices are directly measurable, while others are more subjective and require various means for mapping onto objective parameters.
5
Conclusion and future research
In this paper some important TINA enterprise viewpoint issues have been highlighted. A further enhancement of this viewpoint is proposed with current relevant concepts from the emerging ODP standards. Such an enhancement is then used to analyse the telecommunication service provision process. This analysis has emphasized the fact that one of the most important enterprise issues in today’s open information market is related to QoS concerns. We have then identified some new TINA QoS requirements, followed by a specific focus on the QoS metrics. Finally, we have proposed an applicationindependent QoS metrics framework/methodology, which has been illustrated by applying it to specific service examples. It is our belief that this framework and methodology can provide a useful tool when addressing important QoS issues in the TINA context. There appear to be many issues which need to be addressed in order to respond to demanding techno-economical forces of TINA-like environments. A problem which requires further investigation is associated with the appropriate mapping of a customer’s subjective level of satisfaction to objective indicators. In addition, there is a need to identify methodologies and provide tools which will assist in expressing a total QoS measure as a function of individual parameters. Another class of problems which are currently under study relate to new economicdriven characteristics of the TINA platforms. Specific characteristics which deserve a closer look are [11]: • the economic nature of QoS issues, which emerges as a result of more access to the public, implying support for a broad range of customers with different preferences. The preferences are normally expressed in terms of theQoS which they require, and the price they are ready to pay. The QoS is directly connected with pricing, which should provide various incentives to users. • the impact of the level of trust among agents in the service provision process • the existence of uncertainty which arises due to openness, distance and separation.
Acknowledgements The author would like to thank to Mirion Bearman, Andrew Lister and Barry Kitson for their valuable comments and suggestions on an earlier version of this paper. The work reported in this paper has been funded in part by the Cooperative Research Centres Program through the department of the Prime Minister and Cabinet of the Commonwealth Government of Australia.
References [1] Barr, W., Boyd T., and Inoue Y., ‘The TINA Initiative’, IEEE Communications Magazine. March 1993. [2] Barr, W., Boyd T., and Post M., ‘The Telecommunication Information Networking Architecture Initiative’, Proceedings of the IFIP TC6/WG6.4 International Workshop on Open Distributed Processing, Berlin, Germany, 8-11 October, 1991. [3] ISO/IEC 10746-3 Draft Recommendation X.903: Basic Reference Model of Open Distributed Processing - Part 3: Prescriptive Model, November 1992. [4] Minoli, D., ‘Broadband Network Analysis and Design’, Artech House, 1993. [5] Chabernaud, C., Goerlinger, S., ‘Requirements on IN nodes to meet QoS objectives’, Proceedings of the International Council for Computer Communications Intelligent Network Conference: “Intelligent Networks, the path to global networking”, Tampa, Florida, 1992. [6] Milosevic, Z., Phillips M., ‘Some New Performance Considerations in Open Distributed Environments’, IEEE International Conference on Communications, ICC’93, Geneva, May 1993. [7] OECD Information Computer Communication Policy, ‘Performance indicators for public telecommunications operators’, 1990. [8] Quality of Service: Conceptual Issues and Telecommunications Case Study, Australian Bureau of Transport and Communications Economics, Canberra, 1992. [9] Lancaster, K., ‘Consumer Demand, A New Approach’, Columbia University Press, 1971. [10] Berg, S., Lynch, J. Jr., ‘The measurement and encouragement of telephone service quality’, Telecommunication Policy, April, 1992. [11] Milosevic, Z., Lister, A., Bearman, M., ‘New Economic-driven Aspects of the ODP Enterprise Specification and Related Quality of Service Issues’, International Conference on Open Distributed Processing, ICODP’93, Berlin, Germany, September 1993 (to appear).
Appendix - relevant TINA terminology overview The following terminology is taken from [1]: TINA adopts an object oriented approach, where objects form natural boundaries for distribution. They encapsulate data and processing. A service is a set of capabilities provided by an object, that can be used by other objects. Each service is described by a service interface and service attributes. One or more computational objects form a building block (BB). A BB is built, installed and maintained as a unit and represents a unit of operability, distribution, security and interoperability. TINA adopts a so called segmentation principle, which is a key separation that should be adhered to when grouping objects into BBs. This provides separation of transmission/ switching functionalities (i.e. technology related concerns) from those that are not dependent on these technologies/architectures. These service segment BBs and delivery segment BBs interact with each other only via contracts. A contract is a service provided by an object within BB and which is visible outside that BB. A TINA infrastructure component (i.e. a server within the TINA Distributed Processing Environment, DPE) called contract trader provides contract registration and contract selection services. Namely, servers register their contracts to the contract trader, so that a BB can locate a contract by specifying the contract type and values for the service attributes defined on the contract type.
