![• Prefetching data in batches
@BatchSize(size=10)
- blind-guess optimization
- Batch size = 50
[50, 25, 12, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1]](https://image.slidesharecdn.com/hibernate-performance-tuning-160603214956/85/Hibernate-performance-tuning-25-320.jpg)
Hibernate performance tuning
- 1. Hibernate performance tuning by Igor Dmitriev
- 3. In the beginning • The most difficult to configure an application • Default settings.
- 4. • The huge amount of data stored in database has a great impact on performance. • Customers have major concerns over slow performance of the application.
- 5. What should I do? To grap statistics • Aspect for Service method execution time • P6Spy for SQL logging • High load testing • Be careful : - Testing in stable state - Do a few iterations - DB SQL cache - Second or Query cache
- 8. There are lots of queries
- 10. Settings Hibernate : 5.0.0.Final DB : Postgresql Java : 8
- 13. 1. @OneToOne is not working?
- 15. There are several possible mappings for one-to-one associations • Sharing Primary key - required @PrimaryKeyJoinColumn and @OneToOne(fetch = FetchType.LAZY, optional = false)
- 16. • Lazy loading with bytecode instrumentation • Using a Join Table with mapping table • @OneToMany instead of @OneToOne trick
- 17. @OneToOne • It’s important where FK is. Hibernate has to find out if foreign key exists to create proxy. You can simply make the association unidirectional, keep the link where foreign key resides
- 18. • Do you really need a separate table? - one table for one-to-one mapping - use @Embeddable
- 19. 2. EAGER is not working?
- 21. 3. N+1
- 23. There are a lot of different approaches to resolve N+1 • Prefetching collections with subselects @Fetch(FetchMode.SUBSELECT) - just for collections - Hibernate remembers the original query used to load the elements.
- 24. • Prefetching data in batches @BatchSize(size=10) - blind-guess optimization
- 25. • Prefetching data in batches @BatchSize(size=10) - blind-guess optimization - Batch size = 50 [50, 25, 12, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1]
- 26. • HQL, Criteria or SQL fetch join
- 27. • Hibernate Fetch Profile
- 28. • EntityGraph - JPA 2.1
- 31. • “Every time you think have to map a collection you have to ask yourself why. A mapped persistent collection is a feature, not a requirement.” Christian Bauer Jan 26, 2006
- 32. EclipseLink • @JoinFetch • query.setHint(“eclipselink.join-fetch”,”client.accounts”) • @BatchFetch(type=BatchFetchType.EXISTS) (JOIN,IN) • query.setHint(“eclipselink.batch.type”,”EXISTS”) • query.setHint(“eclipselink.batch”,”client.accounts”)
- 33. Keys
- 34. Batch processing Update or insert huge amount of entity objects • flush + clear • ScrollableResults • StatelessSession - does not have a persistence context cache and does not interact with second and query cache, no dirty checking - all retrieved entities are in detached state - it bypasses any enabled Interceptor or event listeners or event callback methods - no batching - It’s a lower-level abstraction that is much closer to JDBC
- 36. Read-only • HQL constructor Expressions, Criteria Projections, ResultSetTransformer for read- only data • @Immutable • @Transactional(readOnly=true)
- 37. Caching data • Second-level cache (Reference data is an excellent candidate for shared caching)
- 38. • Query cache - You must ask yourself how often you’re going to execute the same query repeatedly, with the same arguments. - Use the query cache in conjunction with the second-level cache !!! - Query.setCacheable(true)
- 39. Bytecode enhancement • @Basic(fetch=FetchType.LAZY) - Blob - @Formula - Dirty checking enhancement Hibernate 5.0+ version
- 41. The most useful recommendation • Control your queries
- 42. Useful references • Hibernate documentation
- 43. • JPA 2.1 specification
- 44. • Java Persistence with Hibernate 2nd edition 2016
- 45. Questions?