Description: At the physical level, traditional RDBMSs store data row by row which is efficient when adding new data and when most/all columns of a table are needed in queries. In some cases, such as OLAP querying where only few out of many available attributes are requested, it is, however, quite inefficient. In such scenarios, so-called column-stores (such as Vectorwise and SAP HANA) can be used instead. They store data column by column at the physical level (at the logical level, they still deal with relations). In that way, the DBMS does not have to read in attributes that are not needed by the user’s query. Further, the columns can be very efficiently compressed and other optimizations are also possible. In this course, column-stores and their performance optimizations will be studied. This will also include the exploitation of modern CPUs as well as main-memory database systems which keep data in main memory to enable very fast access times.
Format: Readings, lectures, and exercises
Prerequisites: A general background in computer science and general familiarity with database management, as can be achieved through an undergraduate database course, is expected. Participants who have taken a graduate database course will benefit from this additional background.
Learning objectives: The goal of the course is to provide knowledge about how column-stores work and when they should be used. In particular, the participants will learn about advanced storage and processing techniques on modern hardware and state-of-the-art in existing systems.

Organizer: Christian Thomsen, Assoc. Professor, chr@cs.aaau.dk

Lecturers: Professor Wolfgang Lehner, Technische Universität Dresden

ECTS: 2

Time: 28 October (afternoon) - 30 October 2014

Place: Aalborg University

Zip code: 9220

City: Aalborg

Number of seats:

Deadline: 13 October 2014