Data management plays a major role in WebSphere Business Monitor.
The database architecture of
WebSphere Business Monitor supports
the following requirements:
- Isolating runtime processing on the datastore from the client access datastore
to maintain the appropriate processing rate
- Being able to perform updates on the client access datastore and still
respond quickly to client queries
- Optimizing the access to historical datastore for analytical and multidimensional
reporting purposes
The usage patterns of the data in the WebSphere Business Monitor databases
differ according to the using component. The data is used by two major components:
the event processor and the client dashboards. This difference in usage makes
it essential to split the database of the event processing from the database
of the dashboards. The data can be further classified into information associated
with a Business Measures-Modell and information
about processing events.
The dashboards display two types of data, recent data and historical instances
data. The number of recent instances is very small compared to the number
of historical instances. The queries performed on the recent instances need
to be extremely fast and must not be affected by the large number of historical
instances. The two types of data have been split into two databases, Runtime
and Historical. To enhance performance, the architecture supports all functions
with the following:
- A database serving as the definitions container for the Business Measures-Modelle.
It also stores information about other databases.
- A database serving as a transactional database and used by the event processor.
- A database that acts as a near real-time analysis database, supporting
analysis queries without affecting the transactional server. It is used by
the dashboards.
- A database that supports multidimensional analysis over the history of
the transactions. It is used by the dashboards to view historical data.
The
WebSphere Business Monitor databases are
divided into four different databases:
- Repository: Stores Business Measures-Modelle and
event definitions. It also stores the schemas, names and host names of State,
Runtime, and Historical databases.
- State: Stores the current state of the running process instances
and the values of the business measures associated with each process instance.
It is used for event processing by WebSphere Business Monitor server.
- Runtime: The
State and Runtime database store pretty much exactly the same information.
The Runtime database differs only by how some data is stored, how recent
the data is, and how long the data stays. The data in the runtime database
stays at least 24 hours longer than it does in the State database. The purpose
of the Runtime database is to allow user to perform near real time analysis
without affecting the event processing done by the WebSphere Business Monitor server.
The Runtime database serves client queries on recent instances. It stores
the runtime information of the business measures group for efficient reporting.
It is used for dashboards viewing.
- Historical database: Stores the information of the completed instances
and the current state of the running instances in a star schema, for historical,
multidimensional reporting. It is used for dashboards viewing.
Two databases store the monitored events and the Adaptive Action Manager
data. These databases are used internally by the
WebSphere Business Monitor.
No information related to process instances or metrics is stored
in them.
- Emitter: Store the events emitted from the engines. The emitters
database tables reside in the engines databases.
- Action catalog: Stores the events that are defined as situations
and actions that the Adaptive Action Manager must perform on them. It is created
during installation.