Database issue? Client issue?

Question

very similar sql statements. if it returns less than 100 rows, it takes 1 second, which is acceptable. But when it returns 1000+ rows, it takes 10+ seconds, which is not acceptable.
My question is, is it the database issue? or because of the network/ client memory that taking too long to show the data on the sql plus?
Since the execution of the sql is fast, I think the bottleneck is on showing the results....how to improve it?
Thanks for your help!
Ken
=======
1099 rows selected.
Elapsed: 00:00:21.12
SQL>
10 rows selected.
Elapsed: 00:00:00.31
SQL>
Edited by: user9511515 on May 2, 2013 8:55 AM

Answer 1

user9511515 wrote:
very similar sql statements. if it returns less than 100 rows, it takes 1 second, which is acceptable. But when it returns 1000+ rows, it takes 10+ seconds, which is not acceptable.Very similar? Not good enough.
As this means that you cannot simply isolate one aspect (number of rows returned), and use that as a comparative benchmark. You have not provided any evidence that the number of rows returned, is the reason for the difference in performance.
Assuming your sql1 and sql2 are on the same data - another contributing factor to the performance difference could be that sql1 hit the disk (and cached data), and that sql2 conveniently hit the cache and not the disk.
A word from he-who-waves-lead-pipe-and-foams-at-the-mouth - the type of comparison you are attempting is almost always fundamentally flawed. Even an identical SQL executed within seconds of one another, will have different elapsed execution times.
Performance tuning is not about comparing process 1 with process 2 and trying to figure out why one is slow and the other fast. Performance tuning is about examining, in detail, the workload of a process. Because if you have no idea WHAT the process is doing, how can determine which parts are slow and which parts can be optimised?