Design Question: huge column is slowing app, what's best way to speedup?

Question

Hello All,
I have a DPL application that is getting a bit heavy due to one column and needs to be tweaked and wanted to ask people's advice on the best practice to speeding it up. I have an entity that is tracking versions of statistical data.
@Entity
public class ResourceVersion {
     @PrimaryKey(sequence = "resource_sequence")
     @XmlAttribute
     private Long id; //internal primary key
     private byte[] data; //the field slowing everything down.
     @SecondaryKey(relate=Relationship.MANY_TO_ONE)
     private String resourceId; //externally visible key.
     private String name; //field that's actually needed.
     //2 more fields that aren't relevant to example.
     //some getters and setters.
The byte[] data field can be several megabytes and is making the application slow as my DAO layer is processing a few megabytes of data and then discarding them for the web layer.
To abstract the problem:
* I have a CRUD application with a single entity.
* My page that lists the entities is getting slow.
* I have a bean with 6 fields.
* I need 3 of the fields for the page that's listing the entities. The other 3 are only used by a tiny subset of the application.
I need to stop retrieving the large data field for pages that don't need it. I need a subset of my bean.
The way I see it, I have 2 options:
1. Figure out a way using the DPL to retrieve a subset of the fields. Is it even possible?
2. Break the bean into 2, one containing the fields needed for listing and one bean that just stores the data.
My assumption is that I have to break the byte[] field into it's own entity. I thought I'd just quickly post to make sure there's not an easier option.
Thanks,
Steven

Answer 1

Hi Steven,
You're correct that BDB does not support partial record read/write, and therefore the best solution is to break the large blob out into a separate entity (or even store it in a file).
See:
http://www.oracle.com/technology/products/berkeley-db/faq/je_faq.html#3
Partial data is not supported at all for the DPL, but as noted in the FAQ the limited support for it in the base API provides very little performance advantage, if any at all.
--mark