We propose to demonstrate HYRISE, a main memory hybrid database system, which automatically partitions tables into vertical partitions consisting of variable numbers of columns based on access patterns to each table. Using an accurate model of cache misses, HYRISE is able to predict the performance of different partitionings, and to automatically select the best partitions using an automated database partitioning algorithm. Our demonstration will show the results of the physical partitioning based on different query workloads, allowing demo attendees to visualize, fine-tune, and modify the partitioning using a GUI. It will then show how the various physical designs affect the query plans and the performance of the database as a whole. Attendees can thus experiment with various physical models, and can grasp the potential of hybrid partitionings, which achieve a 20% to 400% performance improvement over pure all-column or all-row designs on our realistic hybrid workload derived from customer applications.
[1]
David J. DeWitt,et al.
DBMSs on a Modern Processor: Where Does Time Go?
,
1999,
VLDB.
[2]
Marcin Zukowski,et al.
MonetDB/X100: Hyper-Pipelining Query Execution
,
2005,
CIDR.
[3]
Alexander Zeier,et al.
HYRISE - A Main Memory Hybrid Storage Engine
,
2010,
Proc. VLDB Endow..
[4]
Martin L. Kersten,et al.
Database Architecture Optimized for the New Bottleneck: Memory Access
,
1999,
VLDB.
[5]
George Karypis,et al.
Multilevel k-way Partitioning Scheme for Irregular Graphs
,
1998,
J. Parallel Distributed Comput..
[6]
David J. DeWitt,et al.
Materialization Strategies in a Column-Oriented DBMS
,
2007,
2007 IEEE 23rd International Conference on Data Engineering.