Continuous Query Re-Optimization in Stratosphere

• Applicant: Professor Johann-Christoph Freytag, Ph.D.
• Subject Area: Security and Dependability, Operating-, Communication- and Distributed Systems
• Term: from 2010 to 2015
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 132320961

Cloud computing challenges the architecture of existing DBMS in many ways: Dynamic changes in available resources, dynamic changes in resource requirements, and changes in hardware topology while executing programs or queries. These challenges require new approaches to query processing and query optimization when executing complex queries on a Cloud computing platform. Therefore, the proposed project on continuous re-optimization investigates• when and how to change running query execution plans;• decision (cost) models when and how to change running query execution plans;• approaches on how to avoid re-computing already computed results or missing out on results not yet computed;A prototypical implementation of the models and algorithms will demonstrate the continuous re-optimization for execution plans consisting of standard and non-standard relational operators. We see the proposed project in the context of the other four proposals: only if embedded in a Cloud execution environment using abstractions such as map and reduce in combination with an algebraic operator model shall we be able to demonstrate the advantages of our contribution in the context of challenging applications whose processing needs go beyond those for SQL queries on existing architectures.

• DFG Programme: Research Units
• Subproject of: FOR 1306: Stratosphere - Information Management above the Clouds