[an error occurred while processing this directive]
|
 |
IBM Research - Almaden
SIRIUS: Stream Data Management

Traditional OLTP (On Line Transaction Processing) databases provide efficient
mechanisms to capture and maintain the consistent states of database objects.
In such systems it is possible to update the database object and maintain the
modified state as the new consistent state of the object. Pre and post
processing and notification of state change is supported by
database triggers. Most database systems do not exclusively maintain a list of
such state transitions. However, in some cases, a historical log of
transitions is maintained in either database logs or in replication systems,
which could be used to retrieve the state transitions in retrospect.
OLTP databases were designed with a particular query workload in mind that
required evaluation against a consistent state and hence remains
unsuitable for content management systems. New types of query workloads are
being observed in the area of sensor networks, content management systems
and replication networks which are not only interested in the consistent states
of the object, but also in the transitions of state and the correlation
between the transitions.
This represents a fundamental shift in the way the data is managed and
retrieved. State changes from these data sources may be directly consumed by
the applications or may be correlated with similar state changes across a set
of data sources. Such correlations can either be performed by the content
recipient or within the content routing system during propagation.
This research aims to:
- Further the state of the art in databases to support notification
mechanism.
- Enhance the notification mechanism to provide customization of content,
propagation of contents and EDA (Event Driven Architecture) support.
- Provide language and runtime support for specifying and evaluating continuous queries, providing support for streaming data.
- Enhance the database technology to support message correlation.
- Enhance the temporal database support for maintaining the history of events.
- Enhance the content routing system with subscription consolidation.
|
|
|