Comment 5 for bug 618019
Revision history for this message
| Gustavo Niemeyer (niemeyer) wrote Re: OOPS may be underrepresenting storm/sql time | #5 |
© 2004
Canonical Ltd.
•
Terms of use
•
Data privacy
•
Contact Launchpad Support
•
Blog
•
Careers
•
System status
•
0e1f616
(Get the code!)
Whatever the cost of retrieving an object from the database is, it can probably be classified close to O(1). Of course, when you modify the query to not return any objects by using count(), you're stopping not only the DB and ORM retrieval of such objects (which certainly has that fixed cost), but any other processing done with these objects. Given a sufficient number of objects in such a context, any trivial query (SELECT *) will cause the application to time out.
Given that, I see no reason for Storm to be tracking the fetch time, unless there's proper evidence confirming that the time spent doing fetches is significant in time and in variance.