Hi, i found out that some one-to-one relations in the Superclass added some heavy performance penalties. I ve not seen it in the SQL logs though. Perhaps i missed it. Cant comment on the detachment. It seems that its not the problem at all in my case.
--- regards Marc Logemann http://www.logemann.org http://www.logentis.de Am 03.12.2009 um 14:53 schrieb Rick Curtis: > Marc - > > In the case where you call em.find(...) and it takes 500ms how long does > that operation take if the result doesn't need to be detached? The reason I > ask is that I've been doing some performance work and we found that in some > instances we are spending way too much time detaching Entities. > > -- > Thanks, > Rick > > On Thu, Dec 3, 2009 at 6:06 AM, Marc Logemann <[email protected]> wrote: > >> Hi, >> >> i am facing some weird performance issues. Given this JPA Query: >> >> public class SomeDao { >> @SuppressWarnings("unchecked") >> public Object getMetaData(Foo.class, long oid) { >> Object o = em.find(clazz, client); >> return o; >> } >> .. >> } >> >> >> @Entity >> @VersionColumn(name = "jpaversion") >> @Table(name = "foo") >> public class Foo extends BasicClientMetaData { >> >> @Id >> @Column(name = "client_oid") >> private long client; >> ... >> } >> >> @MappedSuperclass >> public class BasicClientMetaData implements MinMaxWeightAware { >> >> >> @Column(name = "custnr") >> private String customerNumber; >> >> @Column(name = "maxweight") >> private float maxWeightStandard; >> .. >> } >> >> When i call the getMetaData() of SomeDao.class (which is transactional and >> gets detached when its returned to the caller), i measured 500ms for this >> operation. The underlying SQL only consumes 20ms which i checked by setting >> the SQL log level to TRACE. >> >> Why is there such a difference? I mean, detaching cant be such an issue >> from performance perspective. I only have this issue with exactly these kind >> of classes (children of BasicClientMetaData). >> >> But i also have other classes which have the @MappedSuperclass which work >> without performance problems. Does anyone have a hint where i can look out >> for? >> >> >> thanks. >> >> --- >> regards >> Marc Logemann >> http://www.logemann.org >> http://www.logentis.de >> >> >> >> >>
