Management of Temporal Data
Overview
Temporal data management is a critical feature in most data management systems today. Instead of “update-in-place”, modern systems create a new version of an object. Keeping old versions has been become more and more affordable due to sharply falling cost of storage.
Once the cost of keeping these additional versions has been paid, users expect rich capabilities to query and process that data in several different ways: Users wish to compare the current status of their data with the status AS OF a year ago; querying a historical version of the data is typically referred to as time travel. Another example is temporal aggregation, e.g. the analysis of how summary statistics of data are changing over time.
After years of slow adoption, more and more applications require support for extensive, high-performance support in their data management systems.
The purpose of this project is explore methods and data structures to combine the benefits of scalable, high-performance in-memory data bases and expressive temporal data management models and operations.
Project Members and Collobarators
ETH Zurich, Systems Group
SAP, Database Group
- Franz Färber
- Norman May
Uni Freiburg, Web Science Group
Publications
- Martin Kaufmann, Amin A. Manjili, Peter M. Fischer, Donald Kossmann, Franz Färber, Norman May: Timeline Index: A Unified Data Structure for Processing Queries on Temporal Data, ACM SIGMOD 2013, to appear
- Martin Kaufmann, Peter M. Fischery, Donald Kossmann, Norman May: A Generic Database Benchmarking Service, Proc. of the 29th Int'l Conference on Data Engineering (ICDE), Demonstration Track. to appear