Transaction service for ICE

[Martin]

We are looking into ICE as a replacement to our CORBA infrastructure, however, there is one thing that we need more than probably any other service that - and that's the transaction service. Is there any efforts being made to develop a transaction service for ICE yet? I would like to know this very much. Thank you.

Martin

[marc]

At present we do not intend to implement a transaction service for Ice. So far there has not been enough demand for such a service.

[Sebastian Kanka]

Regarding the transaction service:

May be the "persistance service" with Freeze is a good feature to simulate or support a transaction service ?

[marc]

Freeze is fully transactional. However, a transaction service is usually for distributed transactions, independent of databases. Freeze could certainly be used to build such a service.

[Martin]

Hi Marc - sorry to butt in again, I was just curious - is the Opengroup's 'transaction service' spec (if I remember correctly, they are the ones that defined the standard) independent of CORBA or is it dependent on CORBA (or the ORB, rather...)? The notion of transactions needn't necessarily be tied to a particular technology, and the reason I ask this - can the Opengroup's transaction standard (API) be fully implemented 'on top of' ICE?

Sorry if this sounds kind of redundant or obvious, but just wanted to hear your opinion. I am impressed with the clarity of thought that went into ICE, and after reading a Usenet diatribe on this issue (which I accidentaly found out when I was looking for some CORBA info ) along with some info from your site , I am fairly certain that based on technical merit alone (not necessarily market momentum/adoption), ICE wins over CORBA hands down...

[matthew]

Originally posted by Martin
Hi Marc - sorry to butt in again, I was just curious - is the Opengroup's 'transaction service' spec (if I remember correctly, they are the ones that defined the standard) independent of CORBA or is it dependent on CORBA (or the ORB, rather...)? The notion of transactions needn't necessarily be tied to a particular technology, and the reason I ask this - can the Opengroup's transaction standard (API) be fully implemented 'on top of' ICE?

Sorry if this sounds kind of redundant or obvious, but just wanted to hear your opinion. I am impressed with the clarity of thought that went into ICE, and after reading a Usenet diatribe on this issue (which I accidentaly found out when I was looking for some CORBA info ) along with some info from your site , I am fairly certain that based on technical merit alone (not necessarily market momentum/adoption), ICE wins over CORBA hands down...

The opengroup defined the XA specification which provides a standard API to a database which can be used to implement distributed transactions. XA itself is independant (in theory) of the underlying database, and independent of any distributed messaging technology (such as OMG CORBA and the OTS).

It is typically used by a TP vendor to tie the TP processing to the underlying database -- that is TP uses XA to control the transactions (start, end, commit, abort, etc). So, therefore, XA would not be implemented 'on top' of Ice, but rather the other way around.

Regards, Matthew