Startseite :: Kontakt :: Sitemap ::


:: Datenbankseminar - HWS '09


:: Vorlesung

Dozent:wechselt jede Woche
Semester:Herbst-/Wintersemester '09
Termine und Räume:Fr: 13:45 Uhr, Raum C1.01
Erstmals am:11.09.2009
:: Kommentar 
  • Voraussetzungen: Je nach Thema, zumindest jedoch Grundkenntnisse in Informatik.
  • Inhalt:

    With the increasing amount of automation in many areas of business, notably in manufacturing and e-commerce, the need to process large amounts of data in real-time has increased dramatically. Large data streams produced by sensors or generated by system transactions have to be analyzed and suitable reactions have to be taken to ensure seamless operation of the business. Typical examples are the following:

    • In large web-sites, the interactions of users with the interface is logged and the logs are analyzed to support performance monitoring and personalization
    • Credit Card companies monitor purchases by their credit card holders to detect credit card frauds.
    • large scale communication networks require continuous monitoring for detecting bottlenecks or diagnosing network attacks.
    • Fully automated production lines are equipped with a multitude of sensors for monitoring different process parameters. This information is used for detecting and diagnosing breakdowns and for planning maintenance.

    Processing the data streams generated in these applications comes with some specific requirements on the data management technology: Streaming data comes with a notion of time or at least sequence that carries important information. In traditional data management systems usually the data is static and the queries vary. In stream data management systems the queries often stay the same while the data changes frequently, even during the execution of the query. These and other specifics of stream data management have to be taken into account when designing a data management system.

    In this seminar we will take a closer look at techniques and systems specialized in processing streaming data. For this purpose, we have selected a number of topics covering different aspects of stream data processing as well as a number of typical systems that will be presented by the participants during the seminar. More specifically, participants will be assigned to topics. For each topic, we will provide an initial paper for getting started. Based on this paper and additional literature, participants are expected to prepare an oral presentation as well as a written report summarizing the topic under consideration. A more detailed introduction to the topic of the seminar as well as a brief introduction of the topics as well as the assignment of topics to participants will happen in the first week of the term.

    Seminar Topics

    1. The CQL continuous query language
    2. Query Execution and Optimization
    3. Filtering, Punctuation, Windows and Synopses
    4. Sliding Window Queries over Data Streams
    5. XML and Data Streams
    6. Time Series Queries in Data Stream Management Systems
    7. Linear Road: A Stream Data Management Benchmark
    8. The Borealis Stream Processing Engine
    9. Cayuga: A General Purpose Event Monitoring System
    10. Cougar: Query Processing in Sensor Networks
    11. TinyDB: An Acquisitional Query Processing System for Sensor Networks
    12. SwissQM: Next Generation Data Processing in Sensor Networks
  • Archiv:

:: Organisatorisches

  • Die Vorbesprechung zum Seminar findet am ersten Termin im Semester statt.
  • Hinweise zum Halten eines Vortrags:
  • Ein Vortrag sollte ca. 20 Minuten dauern (damit noch Zeit für Diskussionen bleibt).
  • Seminarteilnehmer sollten 2 Wochen vor Ihrem Vortrag Ihre Folien mit Ihrem Betreuer besprechen.
  • Die Folien zu einem Vortrag sollten mindestens eine Woche vor dem eigentlichen Vortrag (möglichst in elektronischer Form) beim zuständigen Mitarbeiter vorliegen.
  • Die einzelnen Seminartermine werden von Frau Seeger verwaltet (
  • Teilnehmer des Seminars sollten sich in der Mailingliste anmelden.
  • Eine Latex-Vorlage für die Seminarausarbeitung befindet sich hier: seminar.tex, literatur.bib, und ein Makefile zum erstellen.
:: Folien der Vorträge (nur intern zugreifbar)  
  • Yu Hao (Betreuer: Rim Helaoui) - The CQL continuous query language - 06.11.2009
  • Fabian Bauer (Betreuer: Pit Fender) - Query Execution and Optimization - 06.11.2009
  • Michael Magin (Betreuer: Rim Helaoui) - Filtering, Punctuation, Windows and Synopses - 13.11.2009
  • Torben Dittrich (Betreuer: Alexander Böhm) - Sliding Window Queries over Data Streams - 13.11.2009
  • Serge Thiery (Betreuer: Rim Helaoui) - XML and Data Streams - 20.11.2009
  • Matthias Sattler (Betreuer: Rim Helaoui) - Time Series Queries in Data Stream Management Systems - 20.11.2009
  • Nikolas Schmitt (Betreuer: Pit Fender) - Linear Road: A Stream Data Management Benchmark - 27.11.2009
  • Raoul Feukia Tengomo (Betreuer: Alexander Böhm) - The Borealis Stream Processing Engine - 27.11.2009