Data Structures for In-Memory Databases (Sommersemester 2012)

Dozent: Prof. Dr. Hasso Plattner (Enterprise Platform and Integration Concepts) , Johannes Wust (Enterprise Platform and Integration Concepts)

Allgemeine Information

• Semesterwochenstunden: 4
• ECTS: 6
• Benotet: Ja
• Einschreibefrist: 1.4.2012 - 25.4.2012
• Lehrform: Projektseminar
• Belegungsart: Wahlpflichtmodul

Studiengänge, Modulgruppen & Module

Beschreibung

This seminar covers advanced database topics in the context of in-memory databases with regards to enterprise data management. At the beginning of the term the students will be provided with the necessary knowledge in a short series of lectures. The main part of the seminar will consist of individually preparing a topic in close supervision with a lecturer.

The following topics will be covered:

• Analytical model for In-Memory Database Operators
• Merge Process Optimizations (In-place / Rebuild)
• Strategies for a Merge Process
• Merge Process Scheduling in Mixed Workload Environments
• Bit Compression for Dictionary Encoding in In-Memory Databases
• Impact of Data Distribution on Data Processing
• Impact of the Merge Process on the Transactional Behavior
• Inverted Index Maintenance
• In-Memory Databases on Mobile Devices
• Leveraging GPU's for In-Memory Databases

Voraussetzungen

• Completed Bachelor studies
• Participation in courses Datenbanksysteme I and II is helpful
• C++ programming skills are desirable.

Leistungserfassung

Students have to deliver all of the items below:

• Presentations (all graded)
  • Intermediate presentation (10 mins talk + 5 mins discussion)
  • Final presentation (20 mins talk + 10 mins discussion)
• Testbed for measurements
• Paper (12 pages in ACM two-column style)
  • an intermediate version for reviewing (required)
  • final version (graded)
• 1 paper review (a colleague’s paper) (graded)

Zurück