Subject | RE: [IBO] IBO seems slower for some things compared with the BDE |
---|---|
Author | Peter McLeod |
Post date | 2003-06-27T04:42:01Z |
Hello Helen,
Thank-you for your reply and sorry about the attachment. I will only attach
relevant snippets in future.
Referring back to your earlier advice the application that I am porting, doesn't
have a gazillion active datasets. Most datasets are created dynamically and
then destroyed when fished, in order to reduce network traffic and conserve
system resources. Where ever possible I try and use appropriate client server
design techniques (For instance refer to my article on Delphi 3000. Improve
Interbase Client Server Performance. I can't always do this as sometimes the
Boss, prefers things done in a different fashion, but this is not the case here).
In any instance, I would still imagine that IBO would be faster than BDE as IBO
talks direct to the underlying API to achieve it's result.
I am performing the current test of IBO compared to BDE on a 100 Meg network,
so bandwidth shouldn't be a problem (so I have left SchemaCacheDir as nil,
although I experimented with it following your e-mail, and essentially found
it no effect on performance in this instance). The connections have been tested
to ensure that they are functioning correctly. Also I have uninstalled the bde
to test IBO (and make sure it has been removed from the application), and then
re-installed it afterwards to compare back against IBO. This ensures that the
performance tests are done in the same environment. Both systems (IBO and BDE)
connect to our test server with an Interbase database to ensure test results
should be indicative.
I also performed the test locally, and found that the BDE was faster than IBO.
Do you have any other suggestions for things to look at?
Regards
Peter Mc Leod
Thank-you for your reply and sorry about the attachment. I will only attach
relevant snippets in future.
Referring back to your earlier advice the application that I am porting, doesn't
have a gazillion active datasets. Most datasets are created dynamically and
then destroyed when fished, in order to reduce network traffic and conserve
system resources. Where ever possible I try and use appropriate client server
design techniques (For instance refer to my article on Delphi 3000. Improve
Interbase Client Server Performance. I can't always do this as sometimes the
Boss, prefers things done in a different fashion, but this is not the case here).
In any instance, I would still imagine that IBO would be faster than BDE as IBO
talks direct to the underlying API to achieve it's result.
I am performing the current test of IBO compared to BDE on a 100 Meg network,
so bandwidth shouldn't be a problem (so I have left SchemaCacheDir as nil,
although I experimented with it following your e-mail, and essentially found
it no effect on performance in this instance). The connections have been tested
to ensure that they are functioning correctly. Also I have uninstalled the bde
to test IBO (and make sure it has been removed from the application), and then
re-installed it afterwards to compare back against IBO. This ensures that the
performance tests are done in the same environment. Both systems (IBO and BDE)
connect to our test server with an Interbase database to ensure test results
should be indicative.
I also performed the test locally, and found that the BDE was faster than IBO.
Do you have any other suggestions for things to look at?
Regards
Peter Mc Leod