Seminar: Columnar Databases

58312305
3
Software Systems
Advanced studies
A column-oriented DBMS is a database management system (DBMS) that stores data tables as sections of columns of data rather than as rows of data, like most relational DBMSs. This has advantages for data warehouses, customer relationship management (CRM) systems, and library card catalogs, and other ad-hoc inquiry systems[1] where aggregates are computed over large numbers of similar data items. It is possible to achieve some of the benefits of column-oriented and row-oriented organization with any DBMSs. By denoting one as column-oriented, we are referring to both the ease of expression of a column-oriented structure and the focus on optimizations for column-oriented workloads. This approach is in contrast to row-oriented or row store databases and with correlation databases, which use a value-based storage structure. This seminar examines different column-oriented database systems and what kind of benefits column-oriented databases provide for a different kind of workloads.
Year Semester Date Period Language In charge
2012 autumn 07.09-07.12. 1-2 English Jan Lindström

Lectures

Time Room Lecturer Date
Fri 8-10 C220 Jan Lindström 07.09.2012-07.09.2012
Fri 8-20 C220 Jan Lindström 07.12.2012-07.12.2012

Information for international students

Working language for this seminar is English.

Contact lecturer by email: jan. lindstrom. (at).cs.helsinki.fi or jan.lindstrom. (at). fi.ibm.com

 

General

A column-oriented DBMS is a database management system (DBMS) that stores data tables as sections of columns of data rather than as rows of data, like most relational DBMSs. This has advantages for data warehouses, customer relationship management (CRM) systems, and library card catalogs, and other ad-hoc inquiry systems where aggregates are computed over large numbers of similar data items.

It is possible to achieve some of the benefits of column-oriented and row-oriented organization with any DBMSs. By denoting one as column-oriented, we are referring to both the ease of expression of a column-oriented structure and the focus on optimizations for column-oriented workloads. This approach is in contrast to row-oriented or row store databases and with correlation databases, which use a value-based storage structure.

This seminar examines different column-oriented database systems and what kind of benefits column-oriented databases provide for a different kind of workloads. Seminar will be organized similarly to conferences i.e. in the first session the topics and timetable are agreed. After this students will prepare their papers and these are refereed at least by two students and lecturer. Finally, oral presentations are given on a two day conferense session.


Deadlines

  • Final paper submission: 25.11.
  • Presentation material: 29.11 (PDF, PowerPoint, or Symphony)
  • Workshop day 1 for presenting the papers  30.11
    • presentations in conference style; each presenter is reserved a 40 min slot that will consist of 30 min talk,

Place

IBM Helsinki, Laajalahdentie 23

  • Please arrive arround 8:30

Agenda

  • 09.00 -- 09.20 Jan Lindström: Opening
  • 09.20 -- 09.50 Sezin Yaman: Introduction to column oriented database systems: Paper Slides
  • 10.00 -- 10.30 Liliya Rudko: Column Oriented Database Systems: Paper Slides
  • 10.40 -- 11.10 Sami Sorjonen: Using Column-Oriented approach with multidimensional data: Paper Slides pdf Slides powerpoint
  • 11.20 -- 11.50 Arto Kärki: In-Memory Columnar Databases: Hyper:Paper Slides
  • 12.00 -- 12.45 Lunch
  • 13.00 -- 13.30 Kalle Kärkkäinen: Parallelism in Database Operations: Paper Slides
  • 13.40 -- 14.10 Atte Hinkka: Query Execution on Columnar Databases using Compression: Paper Slides
  • 14.15 -- 14.30 Coffee break
  • 14.30 -- 15.00 Weixiong Rao: MonetDB and its application for IR Searches: Paper Slides:
  • 15.10 -- 15.30 Peitsa Lähteenmäki: MapReduce with Columnar Storage: Paper Slides

Links

DBLP search engine: http://www.dblp.org/search/index.php