Cached statements statistics

Question

For reading data with oracle 9.2 thin JDBC drived I got following statistics
while reading lots of datya (lots of smal reads on m:n as two 1:n):
Kodo connection pooling : 14sec
Third party conn pooling with no statement cahing: 22sec
Third party conn pooling with oracle statement cahing: 17sec.
(I wrapped OracleConnectionPoolDataSource and set pool size on each
getPooledConnection() call)

Answer 1

Some more statistics:
Kodo 2.2.3 for the same op with 3-party pooling produced exactly the same
result as 2.4 with no statement caching - 22sec
So statement caching gives pretty good performance boost but real speed
could be obtained if Kodo can implement some SQL optimization like batch
reads (ie you have object A which references B, you read 100 of As with one
SQL but then execute hundred of reads to read 100 of Bs. batch read allows
to join A and B in one SQL and read them both at once) Kodo 2.5 :-)
One particular very common case where batch reading would shine is M:N with
attributes (often implemented as two 1:N)
But my priority is with better concurrency - multiple JDO_LOCKs especially
for collections.
"Alex Roytman" <[email protected]> wrote in message
news:aodj57$oq5$[email protected]..
For reading data with oracle 9.2 thin JDBC drived I got followingstatistics
while reading lots of datya (lots of smal reads on m:n as two 1:n):
Kodo connection pooling : 14sec
Third party conn pooling with no statement cahing: 22sec
Third party conn pooling with oracle statement cahing: 17sec.
(I wrapped OracleConnectionPoolDataSource and set pool size on each
getPooledConnection() call)