From DB Optimizer
Jump to: navigation, search


1. Why should I use a graphical product over scripts?

see: http://sites.google.com/site/embtdbo/graphics-vs-text


Platform specific, for example testing on Oracle it’s less than 1% of one cpu

3. Speed

Sample every second, refresh every 5 seconds

4. Method

Sample once a second user states and build up a model of load on database.
We collect multiple attributes about the active users and the data can be aggregated in different dimensions. We display different dimensions such as top SQL, Event, Session on the database or for a specific session the sessions top sql, the sessions top event, or for a SQL statement the top sessions executing that SQL, the top wait events for the SQL or for an event the top sessions running into that event and the top sql running into that event.

5. Amount of data

Small but concentrated
Amount of data depends on the number or active sessions. More active sessions, more data. Less active sessions less data. Typically data is small for example less than 1M for an hour of profiling for 10 active sessions.

6. Does DB Optimizer use any of the tables or information provided in Oracle's diagnostic pack?

DB Optimizer does not access in any way, shape or form any of the information, tables or views licensed under Oracle’s diagnostic pack.

SQL shows up as "Unknown" in profilers Overiew and Top Activity (Oracle)

The UNKNOWN is one of the following
1. Commit (log file sync)
2. non-sql from background processes
3. uncollected SQL - we gather SQL text every 15 seconds
4. bug(s)
Normally it's 1-3, and you can expand the node and see how much "non-sql" there is which includes commits. On the other hand I've seen cases where we should have gotten SQL and we didn't.
In DB Optimizer 2.5 we will keep retrying missing SQL text, ie if we don't find it the first collect, we will try again.