Crescando

Overview

Crescando is a scalable, distributed relational table implementation designed to perform large numbers of queries and updates with guaranteed access latency and data freshness. To this end, Crescando leverages a number of modern query processing techniques and hardware trends. Specifically, Crescando is based on parallel, collaborative scans in main memory and so-called "query-data" joins known from data-stream processing. While the approach is not always optimal for a given workload, it provides latency and freshness guarantees for all workloads. Thus, Crescando is particularly attractive if the workload is unknown, changing, or involves many different queries.

Project Status

We are building an elastic, fault-tolerant distribution layer based on a novel multicast protocol: E-Cast. We are also implementing dataflow-based SQL query processing on top of Crescando storage.

There are a number of master thesis topics available in the context of Crescando. Please see below for further information.

Project Setup

Crescando is a joint effort of the ETH Zurich Systems Group and Amadeus IT Sophia-Antipolis.

Researchers: Philipp Unterbrunner, Georgios Giannikis

Publications

• Philipp Unterbrunner, Elastic, Reliable, and Robust Storage and Query Processing with Crescando/RB. PhD Dissertation, ETH Nr. 20272, March 2012.
• Philipp Unterbrunner, Gustavo Alonso, Donald Kossmann: E-Cast: Elastic Multicast. Technical Report Nr. 719, ETH Zurich, Department of Computer Science, 2011.
• Georgios Giannikis, Philipp Unterbrunner, Jeremy Meyer, Gustavo Alonso, Dietmar Fauser, Donald Kossmann: Crescando [Demo Paper]. SIGMOD 2010. Indianapolis, USA, 2010.
• Philipp Unterbrunner, Georgios Giannikis, Gustavo Alonso, Dietmar Fauser, Donald Kossmann: Predictable Performance for Unpredictable Workloads. VLDB 2009. Lyon, France, 2009. [Slides]
• Georgios Giannikis: Deadalus, distributed crescando system. Master Thesis, ETH Zurich, Department of Computer Science, Systems Group, 2009.

Available Master Thesis Topics

There are a number of available master thesis topics which would add important functionality and performance improvements to Crescando. Many of these can be scaled down to a semester thesis if desired.

While some of them are more implementation-heavy then others, all topics require a significant amount of original research and creative hacking, and are of broad interest to the emerging fields of scan-based database systems and cloud computing. Please contact Georgios Giannikis if you are interested in working on Crescando.