Re: Making multiple instances of a RMI method in (from?) one server implementation
Gregg Wonderly <[email protected]> Wed, 6 Jun 2007 10:21:57 -0500
| Newsgroups | gmane.comp.java.sun.rmi |
|---|---|
| Message-ID | <[email protected]> |
Niclas Hedhman wrote: > On Wednesday 06 June 2007 21:27, Gregg Wonderly wrote: > >>Niclas Hedhman wrote: >> >>>Very rough sketch, but you should get the idea. >>>It still doesn't solve time-outs and resilience against "processor" >>>failures. >> >>The Jini leasing model and the Jini transaction model provide tools for >>learning of distributed failure and controlling the progress of distributed >>state. > > Yes, as I mention earlier; JavaSpaces is perfect for this kind of problems and > solves the things I point out. But for someone who is 'narrowed down' to one > way of thinking, start looking at Jini/JavaSpaces is a big step for "no > apparent benefits". So, I think it is important to show "Yes, the obvious > solution is simple, but No, you will have a lot of problems to solve and it > will be a *long* time before you got it all right. So, take a look at Jini > first..." Transactions, including the Lease on the transaction life cycle and the participants existance go a long way to solving the majority of timeout and failure detection. Recovering from such events can be a bit more difficult, depending on how you design the work flow, and whether you have distributed global state that needs to be recovered or initialized in fixed/added participants. Jini has the infrastructure to support such systems. It is a toolset though, not a solution to plug in. If you will spend the time to learn about the Jini programming model, and about what things are available, you will spend a lot less time spinning your wheels and rearchitecting/rewriting. Distributed systems take a lot more care and attention to details than single memory space programming does, precisely because of the fact that partial failure is a problem that occurs much more regularly when everyhing is not running in the same process or on the same hardware, where catistrophic failures take out everything, not just one thing. Gregg Wonderly =========================================================================== To unsubscribe, send email to [email protected] and include in the body of the message "signoff RMI-USERS". For general help, send email to [email protected] and include in the body of the message "help". For a list of frequently asked RMI questions please refer to: http://java.sun.com/j2se/1.3/docs/guide/rmi/faq.html To view past RMI-USERS postings, please see: http://archives.java.sun.com/archives/rmi-users.html