Why is database tuning important?
Tuning IMS databases so that they provide the best possible performance for IMS applications is one of the most important responsibilities of IMS database administrators (DBAs).
In the daily use of IMS databases, segments are added, updated, and deleted. When free space is fragmented by a multitude of segment replacements and delete operations, and if segment overflow occurs, the performance of applications can suffer. When available space is used up, new segments cannot be added. When the randomizing logic does not match the distribution of root segment keys or database record lengths, unexpected overflow of segments can occur, and the performance of physical I/O of root segments or database records can suffer. Monitoring and tuning databases is critical to ensure that databases use space efficiently and to prevent performance degradation.
What happens if databases are not tuned regularly?
Free space fragmentation occurs frequently in IMS databases that are used on a regular basis. Having many pieces of fragmented free space that are too small to reuse leads to poor space utilization. Databases that are not as frequently used tend to have less fragmented free space. However, if data is periodically added, available space will eventually be used up. Also, database types that use randomizers, such as HDAM and Data Entry Databases (DEDBs), might have many synonyms that have high multiplicity or an excessive amount of overflow data due to changes in distribution of root keys and the database record profile that was initially projected.
Without appropriate database tuning, lack of space can lead to an unexpected service outage of the database. Also, other circumstances can increase the number of physical I/Os to retrieve data, and cause the performance of applications to suffer. These circumstances eventually result in performance problems.
To prevent such problems, DBAs must proactively maintain their databases by monitoring space utilization on a daily basis, detecting exceptional statuses as early as possible, and establishing tuning plans before problems occur.
How are IMS solution packs beneficial in database monitoring and tuning?
IMS Database Solution Pack, IMS Database Utility Solution, and IMS Fast Path Solution Pack include the Database Sensor component. This component enables DBAs to collect database statistics within their daily operations. The collected database statistics can be kept in a central repository as sensor data and can be accumulated as historical database statistics.
Autonomics Director, an IBM® IMS Tools Base for z/OS® (also referred to as IMS Tools Base) component, schedules and automates policy evaluations of sensor data and notifies DBAs about the status of their databases. The Autonomics Director ISPF user interface simplifies the processes of scheduling and running policy evaluations and of checking monitoring results.
IMS Administration Foundation features of IBM
Unified Management Server for z/OS (also referred to as
IMS Administration Foundation) and IBM Management Console
for IMS and Db2® for z/OS (also referred to as
Management Console) consolidate key IMS
statistics into an intuitive, graphical web interface. Through the web interface, users can view
aggregate information for customizable groups of databases, graphical charts that show
historical sensor data, database exceptions and corresponding recommendations, and the reports
that are generated by IMS Tools.
The combination of IMS
solution packs, IMS Tools Base components, and IMS Administration Foundation or Management
Console greatly simplifies the tasks that are involved in day-to-day database monitoring and
detecting database exceptions. By using these tools, IMS DBAs no longer need to develop and maintain their own database
monitoring programs.
What is policy-based database monitoring and tuning?
As the number of databases within an organization increases, so does the amount of sensor data that is accumulated on a daily basis. This accumulation of data affects the productivity of DBAs, who must expend considerable time and effort analyzing this data.
Policy-based database monitoring and tuning is the ability to associate rules with three different threshold levels (warning, severe, and critical). When a threshold is met, a notification is issued to the DBA, who can then take appropriate measures. For example, analyzing the state of the database and performing free space reorganization. Policy-based monitoring and tuning reduces the amount of time that a DBA must manually monitor and evaluate database statistics.
Policy-based database monitoring and tuning is provided by integrating several components of IMS Tools Base with IMS Database Solution Pack, IMS Database Utility Solution, or with IMS Fast Path Solution Pack. A set of sensor data evaluation rules is included with these solution packs. DBAs can establish a policy-based monitoring and tuning environment simply by specifying thresholds for the three severity levels, establishing an analysis plan for each severity level, and specifying notification methods.
The three different severity levels help DBAs monitor changing trends within their database environment. Lower-level warnings provide the first indication that a database might need attention. Typically, warning signs do not indicate problems that require immediate attention; however, DBAs can act on a warning notification to mitigate a problem before it reaches the severe or critical level.