Going back to JDBC means using Xregistry (Registry which was part of donation). There is nothing bad in that but lets find what are the bottlenecks in current JPA. I am sure we will be able to fix any performance problem by optimizing relationships and configurations.
Thanks Raminder On Sep 15, 2012, at 9:41 AM, Chathuri Wimalasena wrote: > Hi Lahiru, > > Since we have a middle resource layer underneath JPA model layer, it will > not be hard thing to move from JPA to pure JDBC. IMO what we should do is, > try to identify which queries give us performance bottleneck and try to > replace them with pure JDBC. > > But at this point, I do not think it is a good idea to replace everything > from JDBC since we have a complete model implemented. Instead of replacing > everything, we can go for composite model which will have JDBC queries as > well as openJPA queries. > > Regards, > Chathuri > > On Sat, Sep 15, 2012 at 8:51 AM, Amila Jayasekara > <[email protected]>wrote: > >> Hi Lahiru, >> >> [3] suggests some performance improvements on JPA. I am not sure >> whether all those suggestions are applicable to our scenario. It might >> be worth to try them and see the performance improvement, before >> completely giving up on JPA. >> >> [3] >> http://java-persistence-performance.blogspot.com/2011/06/how-to-improve-jpa-performance-by-1825.html >> >> Thanks >> Amila >> >> On Fri, Sep 14, 2012 at 5:55 PM, Amila Jayasekara >> <[email protected]> wrote: >>> You can also find an another thread related to this in [2]. >>> >>> [2] >> http://stackoverflow.com/questions/260618/orm-solutions-jpa-hibernate-vs-jdbc >>> >>> On Fri, Sep 14, 2012 at 5:41 PM, Lahiru Gunathilake <[email protected]> >> wrote: >>>> Hi Devs, >>>> >>>> I have been testing the new registry implementation and I have a feeling >>>> this is getting slow which is the issue that we are trying to address >> for >>>> jackrabbit implementation. This JPA implementation will add an extra >>>> complexity with a big performance overhead. >>>> >>>> I don't think we need to use a framework for this simple repository >> which >>>> has 11 tables with each having less than 6 columns in each table. My >> point >>>> is for a simple database like this, we really don't have to use a >>>> framework[1]. If our system it so damn complex we can live with the >>>> complexities and the performance bottleneck of JPA. >>>> >>>> WDYT ? >>>> >>>> [1]https://forums.oracle.com/forums/thread.jspa?messageID=10567940 >>>> >>>> -- >>>> System Analyst Programmer >>>> PTI Lab >>>> Indiana University >>
