- Database Replication: Data
replication is very attractive
to increase system
throughput, provide fault-tolerance, and have fast local access. We
have been looking at database replication as a middleware layer and as
part of the database kernelt. At the middleware layer, we have
developed Middle-R which runs on top of standard database systems.
Research focus within Middle-R is on various topics such as wide-area
replication, partial replication, relationship between replication
middleware and underlying database system, and data consistency.
Within the database kernel,
we have developed Postgres-R, an extension to PostgreSQL that provides
fast, cluster-based data replication based on the rich semantics of
group communication. Recent work has looked on recovery and
reconfiguration mechanisms for Postgres-R
- McXML: A native XML data management system:
The XML data format
is currently widely used for data exchange between distributed
components and as the common data model to publish data on the web
in bioinformatics information systems). We have developed the native
XML data management system McXML. Its focus is on efficient update
- Exp-DB: A Laboratory Information
The analysis of proteins and their functionality in the cell
involves many complex steps including various types of experiments.
these steps are supported by specific software products resulting in
enormous amounts of data (several Gigabytes per day). Adequate
storage, tracing, and linking of this data is essential. Other
important issues are workflow management, and access control.
- Machine Learning meets Databases.
Many software packages for data mining and machine algorithms assume
all data can reside in main memory limiting the amount of data they can
handle. We are investigating how to plug relational database systems as
storage manager to such learning software. .