I have this class Award mapping the Award table that has a lots of columns(30). When I do list awards I don't need more than few things (id, description) if I use a projection sql-query like below that will do it.

If I use hql that is way too slow because the extra work performed to fetch 30 columns per award.

Is lazy="true" on the field working ? Is there any other (better) way of doing ?

You could create a separate entity - AwardInfo (with less columns) and map it to the same database table.

Please rate if that helped.

I'm a bit afraid of second level cache that's going to cache two entities for same row leading to higher memory consumption.


I have to give it a try. The prev. approach would only have the drawback that I'd have to change some of the queries the rest would stay similar.

Thanks for the reply!!!
Q-2

Hi Q-2,
You can use projection HQL query like below.

HQL : select property1,property2 from Award where ....

don't forget to rate....

so disable 2nd level cache for this particular entity.

True but that is returning an array list of scalars... and no objects!

Another way I found is to do create a Value Object (POJO) that extends Award

AwardSlim extends Award

and then do a HQL

select new AwardSlim( award.id, award.description )
from
where

that will return AwardSlims that are in fact Awards and is quite fast.

Another way would be to detach the little used column into AwardCharacteristics or a AwardDetail table that will be lazy=true but unfortunately that's going to be imply quite a bit of changes...

Hibernate should support column level lazy property. As far as I know it was in 3.0 road map
So all you have to do is to specify lazy="true" for the property.
Normally by default lazy="false" for properties, but can be enabled.
According to some documents
<cite>
Hibernate3 supports the lazy fetching of individual properties. This optimization technique is also known as fetch groups. Please note that this is mostly a marketing feature, as in practice, optimizing row reads is much more important than optimization of column reads. However, only loading some properties of a class might be useful in extreme cases, when legacy tables have hundreds of columns and the data model can not be improved
</cite>
I believe this is right for you case..
Hope this helps

EDIT: But of course.. it is never easy.. you should do bytecode instrumentation for this class.. CGLIB will not be enough *sad*

I did read that info before and is not working, at least not in the way I expected. I need to see an example.

Thanks for writing back!

Hi

You can use criteria with transformer (AliasToBeanResultTransformer)


select onlu fields what you want


Amila

(Don't forget to rate if helps)

I would use a database view which contains only the relevant fields. I think this would be the fasted solution because you delegate the work to the database.

In conjunction with Hibernate Tools the database views are generated just like tables and you have no additional work with mapping a result array to an object or transforming a result to an object.

Is this what you want?

If yes don't use views. you can mapped only selected columns in table. unnessary to create seperate view for selected columns


Amila

(Don't forget to rate if helps)