5 Replies Latest reply on May 8, 2003 10:11 AM by lpienia

    hibernate v castor

    columod

      hi all,
      we are currently starting a project which will have around 500,000 users. the preference is to use clustered jboss with an OR layer.
      do people have any preferences/opinions on the merits of using castor or hibernate.
      cheers,
      colum.

        • 1. Re: hibernate v castor
          weiqingh

          it depends. (of course)

          there is a useful link at
          http://www.rollerweblogger.org/page/roller/20021013


          hibernate can deal with all the development scenario. you have to figure out what category you fall into. (sounds like you are starting new development. much more freedom.)

          i like hibernate better. it has a very nice session model. it relies on jce for server cache, which serves the purpose most of the time. the only thing i didn't like is it doesn't support a standard query language like oql. but you never know if it's important to have a "standard".

          either one you choose, it will out-perform entity beans by far.

          • 2. Re: hibernate v castor
            weiqingh

            forgot to mention, another important feature i like about hibernate is it supports polymorphism query, which in my opinion is really important and lacking in ejb.

            btw, most of the OR frameworks use reflection. hopefully with latest jdk, it won't be an issue.

            • 3. Re: hibernate v castor
              columod

              cheers, that's a very helpful site.
              i also am leaning towards hibernate. even though reflection based OR is going to be slower (but not much slower) i still favour this approach over the jdo approach so reckon i'll go with this.
              as an aside, have you used hibernate in a clustered environment? if so any particular problems you've noticed.
              thanks again for your help
              colum.

              • 4. Re: hibernate v castor
                weiqingh

                no, haven't used hibernate in a clustered env. according to what i read, it would just work fine.

                one potential problem i can forsee is that jcs cache is per jvm. that's why if you read the hibernate doc, they recommend you to use the server cache for read-only objects. i guess they are waiting for jcs to catch up to have a really good distributed cache, which is a hard problem and maybe your app doesn't require a very strict transactional semantic.

                • 5. Re: hibernate v castor
                  lpienia

                  Does anyone have a good example for Hibernate 2.0 with JBoss?

                  Thanks