Hi Ryan, >> The effect of which is to make a transparent >> layer to store data adhering to a common schema on different >> physical databases. The business logic would be totally >> unaware that it's going to different places to get it's data. >> If I write my own ConnectionFactory this would be pretty >> easy to do, however I'd really like to be able to use OJB so >> that we don't waste weeks of time writing O/R classes by >> hand. From what I can tell, OJB doesn't really support what >> I want to do..
> OJB *does* allow to work simultaneously with multiple databases! > It does *not* provide a 2 phase commit mechanism to synchronize > the transactions against all these databases. This must be done > in user code! Of course, if you have a managed environment, JTA would provide 2-phase commit and OJB does work in a managed environment. Raghu. -----Original Message----- From: Mahler Thomas [mailto:[EMAIL PROTECTED]] Sent: Tuesday, January 14, 2003 7:41 AM To: 'OJB Users List' Subject: AW: OJB C/S Hi Ryan, > Hello, > > I have been evaluating OJB for use in our next application > and have a couple questions that I hope someone can answer. > I've already read through the archives (at least those that > matched my query) and saw some disappointing news. However, > I want to see if there may be a way around it. > > Basically we have a very large database and run a large > number of online transactions against it. In the next > version, we plan to 'roll-our-own' DB clustering for > scalability reasons. We like to be able to partition user > records across multiple DB instances based upon a hashing > algorithm. 'Roll-your-own' could be quite a risk in this context! RDBMS like Oracle or DB2 provide native clustering techniques. I would always recommend to use these features! > The effect of which is to make a transparent > layer to store data adhering to a common schema on different > physical databases. The business logic would be totally > unaware that it's going to different places to get it's data. > If I write my own ConnectionFactory this would be pretty > easy to do, however I'd really like to be able to use OJB so > that we don't waste weeks of time writing O/R classes by > hand. From what I can tell, OJB doesn't really support what > I want to do.. OJB *does* allow to work simultaneously with multiple databases! It does *not* provide a 2 phase commit mechanism to synchronize the transactions against all these databases. This must be done in user code! > worse, it appears (from the list archives) > that even the described client/server functionality doesn't > work. Correct c/s mode is currently out of order. But I don't see any reason why you should rely on c/s mode in your application scenario? > Since this is a production system with paying > customers, should I abandon OJB and go for another solution? 1. I don't know of any other o/r tool that does this kind of job 2. If you have paying customers, let them buy a database with built in clustering support! cheers, Thomas > > Thanks, > > Ryan Clifton > Systems Engineering Manager > -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]> -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
