Stream databases Strumieniowe Bazy Danych

Download Report

Transcript Stream databases Strumieniowe Bazy Danych

Stream databases
Strumieniowe Bazy Danych
Przemysław Pawluk
Supervisors:
prof. Zygmunt Mazur (Wroclaw University of Technology)
prof. Lars Lundberg (Blekinge Institute of Technology)
Institute of Applied Informatics
Department of Computer Science and Management
Wroclaw University of Technology
Agenda
• Introduction
• Data model
• Continuous Query Languages
• Results of the work
• Possible usage
• Summary
2/12
Introduction
Goals of the thesis:
• Stream model presentation
• Presentation of complexity assessment
method
• Consideration of pros and cons of the
stream model
3/12
Data model
• Stream data model is an extension of
relational data model.
• Data elements arrived on-line and stay
only for a limited time period
• Data from the stream could be stored as
relation
• The relation could be transformed into the
stream
4/12
Continuous Query Languages
Languages taxonomy
Declarative
extension of SQL
• Project STREAM
• Project TelegraphCQ
Procedural
• Project Aurora
Lack of standard!
5/12
Results of the work (1)
• To consider pros and cons of stream
databases two models have been
compared
• Complexity of models has been assessed
• As a basis of design process the
Hofmeister’s method has been used
6/12
Results of the work (2)
• Models have been transformed into
Weighted Class Dependence Graph
• To enable processing graphs have been
presented in the form of matrix
• The measure of complexity of the model is
entropy
7/12
Results of the work (3)
• Model based on stream database has lower
complexity (7,3%)
• Lower complexity can influence costs of
development and maintenance of the
system
• Usage of stream databases in
telecommunication system could be
profitable
8/12
Results of the work (4)
Issues in the thesis:
• Strong dependency on experience
• Special character of data
• Only academic implementation are
available (problem with comparison of i.e.
performance)
9/12
Possible usage
•
•
•
•
Telecommunication
Medicine
Army
…
Where data can be perceived as a stream
10/12
Summary
• New class of database systems has been
presented
• Method of complexity assessment has been
described
• Two models have been compared in the
context of complexity
• Possibility usage stream databases in
telecommunication systems has been
considered
11/12
Thank you