direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Page Content

Short info

Candidate: Matthias Ringwald

Advisor: Max Heimel

Desired degree: Master

Abstract

Applications usually issue declarative SQL queries via connectors such as JDBC or ODBC to the database. This splits the application into a control flow optimized application part and data flow optimized queries. However, with the rising complexity of data analysis algorithms – paired with growing amounts of input data – a joint optimization of the control and data flow is required in order to allow efficient query execution.

In this talk we present RelL, an imperative data programming language for defining data-intense programs running within SAP HANA databases. By adding relational language constructs to a general-purpose programming language, RelL is expressive enough to allow optimizations known from relational algebra while still being suited for implementing general-purpose applications. By applying join order optimization and utilizing efficient join strategies, the RelL optimizer improves performance by orders of magnitude.

Zusatzinformationen / Extras