• Deutsch (DE-CH-AT)
  • English (United Kingdom)
  • Japanese(JP)
SML@b Metrics Data Base (MDB) PDF Print


The use of metrics in the development of industrial software is gaining importance. Metrics are particularly suited to qualitative and quantitative assessment of the software development process, of the resources used in development and of the software product itself. However, software metrics can only be used effectively if the requisite measurements are integrated into the software development process and if these measurement values are taken at regular intervals. An effective software measurement process produces an extensive series of measurements and thus the need for efficient measurement data management, which must include the contexts to enable discourse on the measurements that are taken as well as to provide extensive evaluation options.

Our industrial approach is published in Ebert/Dumke: Software Measurement - Establish, Extract, Execute and Evaluate, Springer Publ. 2007. The general architecure of metrics data base processes are shown in the following figure.


The involved apects in metrics data base are given in the next figure.


These involvements are in detail:

Database characteristics:

  • kind of used database model (relational, object-relational, hierarchical/XML-based and so on)
  • architecture (layer, kind of possible access, components)
  • user concept (query language, GUI, provided services)
  • security and safety (access control, etc.)
  • quality (consistency, redundancy, performance)

Data characteristics:

  • kind of value (number, flag, text)
  • metadata (precision, tolerance, accuracy, unit, domain)
  • structure and type (aggregation level)
  • source (measurement, statistical operation, default setting)

System characteristics:

  • stand-alone (data analysis, decision support)
  • embedded system (controlling, management)
  • distributed systems (client/server, mobile)
  • Web-based solution (portal, Web services)

Input types:

  • measured value
  • output from a specific tool
  • prediction or estimation

Output types:

  • kind of values
  • input for specific tools
  • statistical report

Metrics characteristics:

  • single measured value (interval, ratio scaled)
  • estimated value (nominal, ordinal scaled)
  • predicted value (by formula, by experience)
  • multi value (tuple, set, tree, tensor)

Measurement characteristics:

  • model-based measurement
  • direct measurement
  • prediction/estimation

Application characteristics:

  • part of management systems
  • source of assessments
  • part of control systems
  • source of education


Our implementation for a OO metrics data base is shown in the following architecture description.


The concept of this MDB was developed from the German Telekom Research Center and the SML@b and the implementation was done by the Bulgarian company Datacom. An application of this measurement data base shows the following screen.




Further general architectures for MDBs are: the measurement data warehouse,


the mediated measurement repository


and the service-bus oriented measurement repository.







Last Updated on Monday, 14 January 2013 14:21