Eclipse Community Forums: EclipseLink

Home » Eclipse Projects » EclipseLink » @cache (size limitations in Eclipselink 2.4.2, JPA 2.0 )

Tue, 02 June 2015 19:17

Eclipse User

Greetings,
My entities are mapped to oracle 12c database. The client needs to check for objects in the incoming stream with the existing ones in the DB. I am doing a getDestDim.Findall() to get the list of result set from DestDim table in the DB.

How can I use cache here and to pre-fetch the result and just query the cache.
Going though the doc, I noticed that @cache has a size property. If I set the size to say 64000.

1.Is this size from the documentation mentioning in bytes ? is it 6.4 MB or 64 MB ?

2. What happens if the records in the db column are more then the size ?
Does the cache grow automatically.

3.When does the cache hit the ceiling and what will happen then ? does fetching of new records fail ? How will the client be notified that the cache is full in all the useless above .

Appreciate your input.

Thanks
Sunny

Re: @cache [message #1697382 is a reply to message #1697291]

Wed, 03 June 2015 10:03

Eclipse User

A cache works by holding a reference to your entities so that they can be accessed again without hitting the database. When it is full, it just means that reading an object that isn't in the cache causes it to be built and placed in the cache - so the least recently accessed one gets pushed out. No warnings or problems should occur, except if you constantly read 10001 records sequentially each time, but have a cache of 10000 it might negate the value of a cache.

Size - the cache deals with your java objects, not their size. Depending on the cache you are referring, 64000 would be the number of references that would be held. There are different caches with different behaviors and different reference types. I assume you have seen https://wiki.eclipse.org/EclipseLink/Examples/JPA/Caching but you should look at http://www.eclipse.org/eclipselink/documentation/2.6/concepts/cache.htm and http://www.eclipse.org/eclipselink/documentation/2.6/solutions/performance001.htm#CHDCGIHF to answer your questions.

Re: @cache [message #1697420 is a reply to message #1697382]

Wed, 03 June 2015 14:40

Eclipse User

Thank you Chris for the reply and the info on the cache. What are the different strategies available to avoid Out of memory errors . I came across

1. pagination
2. Use Coherence cache

In addition I am looking at JPQL to fetch only the required db columns instead of the whole table (with the current impl). Any specific examples on jpql would be helpful

-Sunny

Re: @cache [message #1697658 is a reply to message #1697420]

Fri, 05 June 2015 14:40

Eclipse User

There are a number of options listed at https://wiki.eclipse.org/EclipseLink/Performance and http://www.eclipse.org/eclipselink/documentation/2.5/solutions/performance002.htm that might help, such as the pagination you've mentioned and even read-only queries (since the entities returned don't have to be cloned, it reduces the memory required). Using JPQL to return selected attributes though is handling/returning raw data, so your query may avoid the memory costs of dealing with a full entity, but then you lose the benefits of caching etc. You are getting into tradeoffs that really require understanding your applications design and usage that are hard to understand through a forum.

There are a few simple JPQL examples here: https://wiki.eclipse.org/EclipseLink/UserGuide/JPA/Basic_JPA_Development/Querying/JPQL#Select_query_examples

Previous Topic:	Exception when read Geometry type from Postgis
Next Topic:	eclipselink:session.validate

Goto Forum:

-=] Back to Top [=-

Current Time: Fri Apr 18 06:14:52 EDT 2025

.:: Contact :: Home ::.

Breadcrumbs

Sign up to our Newsletter