4J

For those who have worked with SQLRelay before, you already know the power of this tool. For those who don't, I suggest you have a look at this great database tool.

I have compared the basic performance of the two primary Perl interfaces: 1) the SQLRelay supplied interface and 2) the DBI bindings

Granted - I know that DBI has more overhead, but I wanted to know what the performance difference were.

Results

Here are some RAW results: I have run 10 tests each, first comparing the SQLRelay interface, and then DBI:

SQLRelay:

TEST 1 START
RUN 1 -> time =0.0988240242004395
RUN 2 -> time =0.119891881942749
RUN 3 -> time =0.0670750141143799
RUN 4 -> time =0.0965390205383301
RUN 5 -> time =0.108819961547852
RUN 6 -> time =0.0503008365631104
RUN 7 -> time =0.0784649848937988
RUN 8 -> time =0.113129854202271
RUN 9 -> time =0.0531101226806641
RUN 10 -> time =0.0935280323028564

TEST 1 AVG=0.087968373298645

Next is the DBI test:

TEST 2 START
RUN 1 -> time =0.136901140213013
RUN 2 -> time =0.0799510478973389
RUN 3 -> time =0.0771389007568359
RUN 4 -> time =0.150060892105103
RUN 5 -> time =0.0818459987640381
RUN 6 -> time =0.130953788757324
RUN 7 -> time =0.0769519805908203
RUN 8 -> time =0.129808902740479
RUN 9 -> time =0.0768680572509766
RUN 10 -> time =0.135869026184082

TEST 2 AVG=0.107634973526001

Conclusions

The tests confirm that using the native SQLRelay Perl interface does provide a considerable performance advantage over the DBI interface. However, the DBI interface is much more portable.