· Position
  · Future
  · Papers
  · Confs
  · Links
 
  R   O   D   A   I   N    Position Statement

    Premises

    Databases will already in the near future have a central role in telecommunications networks. The information needed in operations and management of the nets will be collected into a logically uniform database. The world-wide nature of telecommunications prescribes that the only possibility to obtain the logical uniformity is the co-operation of autonomous databases.

    During the next ten years great changes are expected in the ways of doing business on the telecommunications markets. The relative importance of traditional transmission and switching services will decrease. This is due both to liberalisation of telecommunications and to developments in the technology. We anticipate that in the future transmission and switching are done using generic technology based on international standards.

    The ability of a telco operator to survive in the international competition will base either on huge volume of the few bulk services or on enhanced services. The ability of introducing enhanced services will primarily be based on the competence of responding in time to the needs on the markets. Future telecommunication architectures like IN and TINA are developed to meet the goal of rapid and efficient creation, deployment, and management of new services. An efficient and reliable database system is one of the key factors in meeting those goals.

    Most of the recommendations issued in the 90's by ITU-T and other standardisation bodies are object oriented. Therefore, object orientation is a natural starting point for the telecommunications database system, too. However, object-oriented databases are still emerging. They are primarily designed for situations where objects have complex structures and transactions are long-living. In telecommunications the situations are different. Most objects have quite simple structures and transactions are short. In addition, we anticipate that real-time properties, guaranteed response time in particular, are soon needed.

    Due to the world-wide nature of telecommunications international standards - both official and de facto - are mandatory. In this project we have identified the following standards as our prerequisites:

    • IN recommendations by ITU-T and ETSI
    • TINA architecture by Tina Consortium
    • TMN by ITU-T and ETSI
    • RM-ODP by ISO and ITU-T
    • CORBA by Object Management Group (OMG)
    • ODMG-93 by Object Database Management Group (ODMG)

    Objectives

    In the research project Rodain we continue the work done in the Darfin project (Database Architecture for Intelligent Networks) funded by Telecom Finland. The ultimate objective of the Rodain project is to design and to specify a real-time object-oriented database architecture for Intelligent Networks and to implement a prototype based on that architecture. The design and specification take place in 1996 while the prototype will be implemented in 1997.

    In particular, we will carry out in 1996 the following research activities:

    1. Evaluation of Concurrency Control Algorithms for Real-Time Databases
    2. Specification of Data Models for Real-Time Databases in Telecommunications
    3. Evaluation of Platforms for Real-Time-Databases
    4. Specification of RODAIN Database Architecture and Interfaces

    Partners

    The Rodain project is a collaborative enterprise of Nokia Telecommunications, Solid Information Technology, Systems Software Partners, Telecom Finland, and Department of Computer Science in University of Helsinki. The Rodain project is partially funded by the Finnish Technology Development Centre (TEKES).

    Darfin Inheritance and Rodain Directions

    The results of the Darfin project are published in several public reports and international publications (see References). Below we briefly summarise the key results and their implications to the research to be done in the Rodain project.

    High-Level Information Model

    On an abstract level we can interpret that a telephone call is controlled by call processing logic and data. The data can be termed as Call Processing Data Object. It is a dynamic object that is created during the call establishment. During the call the object may change as a result of the so-called mid-call service control actions. Logically the data object can be divided into two parts: Calling Party Object and Called Party Object. Moreover, the data object may have attributes that trigger actions to modify and/or to create persistent data objects when the call is terminated.

    Figure 1 summarises our abstract information model of a telephone call. It should be noted that the call is not necessarily between two telephones. The called party may also be an information processing system. User management of services is a typical case in which the called party is an information processing system.

    Figure 1: A High-Level Information Model of Phone Call

    The Call Processing Data Object is typically created in two phases. The first phase consists of creating the Calling Party Object. When the switch detects that the hook is lifted up, a query is triggered. The query builds up the object by retrieving the service profile for the originating line from the persistent subscriber data object.

    The Called Party Object is created in the second phase. This object depends on the access number dialled. Typically, the dialled digits are given as the key in the query that creates the Called Party Object. The query builds up the object by retrieving the service profile for the destination line from the persistent subscriber data object.

    Real-Time Transactions

    Based on the IN Recommendations of ITU-T (Q.1204 and Q.1211) we have concluded that real-time transaction processing capabilities will be needed. Real-Time Database System (RT-DBS) is a transaction processing system that tries to satisfy the explicit timing constraints (deadlines) associated with each incoming transaction. The ultimate goal of any RT-DBS is to maximise the fraction of transactions that meet their deadline. This is very different from the goal of traditional database management systems (DBMSs) that try to minimise the average response time.

    When a database system for telecommunication services is to be designed, the list of most important issues includes real-time. We anticipate that both firm and soft deadlines will be needed. When the deadline is firm (outcome of a late transaction is useless or harmful), then the late transaction (or transaction that can no more meet its deadline) can be aborted without restart. When the deadline is soft (outcome of a late transaction gradually loses its value), then the late transaction is continued with a lower priority. We do not believe that hard (meeting the deadline must be guaranteed) deadlines are needed in telecommunications applications.

    One of the primary difficulties in designing an RT-DBS is the fact that the concurrency control of data access must be combined with real-time scheduling. The major design issues [see i.e. Raatikainen and Taina 1995; Yu et al. 1994] in concurrency control are conflict detection and conflict resolution. Of the algorithms proposed in the literature our current favourite is the 2PL-HP algorithm [Haritsa et al. 1992]. In that algorithm a transaction having a higher priority aborts the transactions with lower priorities when a lock conflict occurs. The transactions with lower priorities are later restarted if the system still estimates that their deadlines will not expire. However, the Rodain project will carry out an evaluation study of the concurrency control algorithms for real-time databases.

    In the Darfin project we have learned the basic understanding of real time, object orientation, and database needs in telecommunications. In particular, the simple prototype implemented gave us valuable experience on the problem space. The database architecture should be flexible due to different kinds of needs present in telecommunications; for details see Taina [1995]. Based on the experience we believe that the easiest way to implement the required functionality is to use an object-based approach. An open question is if a fully object-oriented architecture - something similar to ODMG-93 [Cattell 1994] - is really needed. Currently we believe that it is better to use a reduced set of object features than to use a fully object-oriented database when we are developing a real-time database to be used in telecommunications. In Rodain we will start from a real-time database system and include the necessary object features.

    Object Models

    In the Darfin project we evaluated four different object models: X.500, X.700, OMG, and ODMG; for details see Raatikainen et al. [1995].

    Of those models the X.500 model is the most mature. However, there are some major shortcomings in X.500. The two most serious ones are the formal organisation of data into a directed graph and the lack of atomic transactions. Minor shortcomings include the lack of multiple inheritance, operations, relationships, extents, keys, and transient objects.

    The formal organisation as a directed graph makes it complicated to use X.500 as the object model of the database system for telecommunication services. As pointed out in the usage scenario, the queries have different criteria in the select operation: the line number and the dialed digits. Moreover, the management needs at least two different views into the data: one in which all services of a single subscriber can be found and another in which all subscribers of a single service can be found.

    The Management Information Model defined in the X.700-series of ITU-T Recommendations is a full object model. Although a distributed management system as defined in X.700 is actually quite close to a database management system, most drawbacks identified in X.500 are also present in the X.700 Information Model. In particular, it is stated in the X.700 recommendations that the local databases containing the information are independent from the recommendations. When taking the management's point of view, this is flexible. However, the approach leaves everything open when the point of view is that of an object-oriented database system.

    The ultimate goal in OMG is portability and interoperability. These goals are also very important in telecommunications. However, they are so general that the OMG Core Object Model does not take into account the specific needs of databases. Therefore, our choice as the baseline object model is the ODMG Object Model. However, the OMG - particularly the OMG Object Request Broker - will obviously have an important role in future telecommunication systems.

    The ODMG-93 is designed to be a general-purpose object database system. Therefore, all features presented in the ODMG-93 Release 1.1 [Cattell 1994] are neither appropriate nor needed in a real-time database system; see Taina and Raatikainen [1995]. An ''RT-ODMG'', that is a real-time ODMG, will be a subset of ODMG-93 with predefined types of real-time object and real-time transaction. Type real-time object will include attributes that specify the correctness criteria and resource demands of operations. Type real-time transaction will include attributes that may overwrite the correctness criteria of operations and that specify the deadline and criticality for the instances of this type.

    In Rodain we will develop a data model tailored for the needs of telecommunications. We will examine how the object model of ODMG-93 should be modified to meet those needs. The interfaces of data access will be flexible. In particular, applications using the database services can use INAP as specified in the IN Capability Set 1 of ITU-T (Q.1218), CMIP as specified in X.700, ODP channels as specified in RM-ODP (X.900 series), and CORBA as specified by OMG. In addition, we will provide an Object Query Language (OQL) interface which will be a subset of the ODMG OQL.

    Correctness Criteria for Real-Time Transactions

    Correctness of transaction results in real-time databases is an interesting area of research. In real-time applications concurrent transactions do not necessarily need serializability [Eswaran et al. 1976] to produce a correct result. In many situations the result of a transaction can be regarded as correct as long as the result corresponds to the real-world situation presented in the database even if the transaction does not serialise with other transactions.

    An interesting correctness criteria called tau-serializability was introduce in Raatikainen et al. [1995]. It can be used to reduce the number of serialisation conflicts. In tau-serializability a transaction is allowed to read old data as long as the update is not older than a predefined time. In the Rodain project we will further refine and examine the basic concept of tau-serializability.

    References

    Cattell, R. G. G. (Ed.). 1994. The Object Database Standard: ODMG-93. Version 1.1 edition. Morgan Kauffmann, San Francisco, Calif.

    Eswaran, K. P., Gray, J., Lorie, R. A. and Traiger, I. L. 1976. The Notions of Consistency and Predicate Locks ib a Database System. Communications of the ACM 19, 11 (November): 624-633.

    Haritsa, J. R., Carey, M. J., and Livney, M. 1992. Data Access Scheduling in Firm Real-Time Database Systems. Journal of Real-Time Systems 4, 3 (September): 203-241.

    Raatikainen, K., Karttunen, T., Martikainen, O., and Taina J. 1995. Evaluation of Database Architectures for Intelligent Networks. Proc. of Telecom95 Technical Summit, Vol 2, 549-553. ITU, Geneva, Switzerland.

    Raatikainen, K. and Taina, J. 1995. Design Issues in Database Systems for Telecommunication Services. Proc. of IFIP TC6 Working Conference on Intelligent Networks. August, Copenhagen, Denmark.

    Taina, J. 1995. Requirements Analysis for Database Services in Telecommunications. Report C-1995-17. Department of Computer Science, University of Helsinki.

    Taina, J. and Raatikainen, K. 1995. Design Issues and Experimental Database Architecture for Telecommunications. Intelligent Networks 1995. Chapman and Hall, London, UK.

    Yu, P. S., Wu, K.-L., and Son, S. 1994. On Real-Time Databases: Concurrency Control and Scheduling. Proceedings of the IEEE 82,1 (January): 140-157.

    Darfin Publications

    Kimmo Raatikainen: Database Access in Intelligent Networks, in Intelligent Networks (Chapman & Hall, 1995), pp.173-193

    Juha Taina: Problem Classes in Intelligent Network Database Design, in Intelligent Networks (Chapman & Hall, 1995), pp.194-207

    Kimmo Raatikainen: Information Aspects of Services and Service Features in IN CS-1, Technical Report C-1994-45, Univ. of Helsinki, Dept. of Computer Science

    Juha Taina: Evaluation of OMG, ODMG, X.500, and X.700 Data Models, Technical Report C-1994-54, Univ. of Helsinki, Dept. of Computer Science

    Kimmo Raatikainen and Juha Taina: Design Issues in Database Systems for Telecommunication Services, Technical Report C-1995-16, Univ. of Helsinki, Dept. of Computer Science. Also in Proc. IFIP IN'95 Conference

    Juha Taina: Requirements Analysis for Database Services in Telecommunications, Technical Report C-1995-17, Univ. of Helsinki, Dept. of Computer Science

    Juha Taina, Tapani Karttunen, Olli Martikainen, and Kimmo Raatikainen: Evaluation of Database Architectures for Intelligent Networks, Technical Report C-1995-29, Univ. of Helsinki, Dept. of Computer Science

    Juha Taina, Mika Rautila, and Kimmo Raatikainen: An Experimental Database Architecture for Intelligent Networks, Technical Report C-1995-30, Univ. of Helsinki, Dept. of Computer Science. Also in Proc. IFIP IN'95 Conference

    Juha Taina and Harri Töhönen: On Implementing a Prototype IN Database Architecture, Technical Report C-1995-42, Univ. of Helsinki, Dept. of Computer Science.

    Mar 27, 1996
    Kimmo.Raatikainen@cs.Helsinki.FI
Last Modified on Oct 06, 1998 by Tiina.Niklander@cs.Helsinki.FI