If an existing environment is to be modified or compared with another, the creation of a load profile from the complete database or the major applications operated, is suitable. The most elegant possibility for doing this is the collection of a comprehensive group of tracefiles.

To detect different behaviour, an actual/theoretical or before/after comparison on the basis of the SQL commands stored in the repository, the related execution plans, the wait events arising and the time elapsed could be executed.

Using a RAC database, the procedures described below can be carried out on the level of the individual nodes, which makes it possible to recognize asymmetries between.

A possible procedure might look as follows:

  1. Tracefiles are selected, either individually as file or as bulk from a directory, and loaded in the repository.
  2. During the load all SQL commands
    a. are brought on a common time basis and
    b. a list of involved instances is created
  3. In the InstanceFilter, instances can be made inactive selectivly. A further limitation can be introduced with the limitation of an interval on the TimeFilter.
  4. In the different DetailTabs, all information on executions of a SQL command which is not active is greyed out. If there is no active information at all for a SQL command, it will also be greyed out in the repository view in addition.
  5. Through the inactive switching of individual instances and the following scrolling through the repository, anomalies between the instances can be recognized. This could be out of
    a. differences in the structure of SQL command itself and/or
    b. different ways of execution.

QueryAdvisor trial.
QueryAdvisor in 30 Seconds

QueryAdvisor quote.
QueryAdvisor support forum.
QueryAdvisor Twitter channel.
social icons Facebook Twitter Blogspot RSS