Subject Jaybird Class 2 & Class 4 remote performance comparison.
Author Ryan Baldwin

I have re-run the tests I run on a local database against a remote database.
Client and server are both win2k and Firebird 1.0.2 was used.

Here are the numbers:

Type 4 Type 2

Run#1 Run#1
benchmarkPreparedInsert: 35111 benchmarkPreparedInsert: 42611
benchmarkNonPreparedInsert: 37163 benchmarkNonPreparedInsert: 45045
benchmarkBigBlobInsert: 57172 benchmarkBigBlobInsert: 65284

Run#2 Run#2
benchmarkPreparedInsert: 34590 benchmarkPreparedInsert: 38926
benchmarkNonPreparedInsert: 36683 benchmarkNonPreparedInsert: 40097
benchmarkBigBlobInsert: 57563 benchmarkBigBlobInsert: 60728

Run#3 Run#3
benchmarkPreparedInsert: 34660 benchmarkPreparedInsert: 38205
benchmarkNonPreparedInsert: 36752 benchmarkNonPreparedInsert: 40308
benchmarkBigBlobInsert: 56562 benchmarkBigBlobInsert: 60997

The big blob tests only used 1k blobs. These numbers seem to show that the
Jaybird type 4 mode has an edge over type 2 mode for remote connections.

Somthing I found kind of interestiing is that for a local connection non
prepared inserts of the same data over and over runs consistantly faster
then inserts using a prepared statement. Its only when a remote connections
is used that prepared statements seem to have a performance advantage for
repeated inserts.