When you create a tracing session, you have the option of storing tracing data within the database being profiled. This is suitable for development environments where you are testing applications, or if there are few connections to the database. However, if your database typically handles 10 or more connections at any given time, it is recommended that you store tracing data in a separate tracing database, in order to have a minimal impact on performance.
The easiest method for creating a separate tracing database is to do so while starting a tracing session using the Database Tracing wizard. The Database Tracing wizard unloads schema and permission information from the production database when creating the tracing database. The tracing database that you create can be used to store data for subsequent tracing sessions as well. For information on how to create a tracing session, see Creating a tracing session.
If you do not want to create a tracing session, but want only to create a tracing database, you can do so manually using the Unload utility (dbunload).
Connect to the database as the DBA.
Execute a dbunload command, similar to the following, to unload the schema from the production database into the new tracing database:
dbunload -c "UID=DBA;PWD=sql;ENG=sample;DBN=sample" -an tracing.db -n -k
This example creates a new database with the name supplied by the -an option (tracing.db). The -n option unloads the schema from the database being profiled (in this case, the SQL Anywhere sample database, demo.db) into the new tracing database. The -k option populates the tracing database with information that the application profiling tool uses to analyze the tracing data.
If you want to store the tracing database on a separate computer, copy it to the new location.