Subject | Re: IB 6.1 Compatibility |
---|---|
Author | Brian Gooch |
Post date | 2010-11-11T14:53:26Z |
Jason,
In the end we decided to take everything in easy stages: going from IB4.xx/6.x to FB2.0/2.5 on 64-bit was just far too much in one go.
1. IB xxx -> FB1.0 all dialect 1 : straightforward. FB runs in the same environment.
2. Modified a copy of FB1.0 to make datatypes match those in dialect 3; data migration to FB1.5 courtesy of a data pump (DWB4) to match columns in tables and migrate in one go.
3. As we do not have the volumes of data whereby we would benefit from 64-bit addressing, we have stayed with 32-bit. We did initially try XP Pro-64bit as a test, but there were many probs with the app - see Evgeniy Rhyzkov paper on 64-bit apps and it is easy to see why there were probs (contact evg@...) - hence another reason to stay 32bit.
4. Still in 32-bit, the app still needed some changes to accommodate the dialect 3 datatypes. Thus, once in FB1.5, we do not see an easy path back to the FB1.0/IB4/6 scenario - that has to be dropped.
5. -> 64-bit and FB2.x, I would suggest is another stage after having ensured FB1.5 is totally stable - shortcutting the stages is not an option. Every step must be safe and secure, robust and reliable.
Do read Evgeniy's paper. HTH
Kind Regards,Brian Gooch
[Non-text portions of this message have been removed]
In the end we decided to take everything in easy stages: going from IB4.xx/6.x to FB2.0/2.5 on 64-bit was just far too much in one go.
1. IB xxx -> FB1.0 all dialect 1 : straightforward. FB runs in the same environment.
2. Modified a copy of FB1.0 to make datatypes match those in dialect 3; data migration to FB1.5 courtesy of a data pump (DWB4) to match columns in tables and migrate in one go.
3. As we do not have the volumes of data whereby we would benefit from 64-bit addressing, we have stayed with 32-bit. We did initially try XP Pro-64bit as a test, but there were many probs with the app - see Evgeniy Rhyzkov paper on 64-bit apps and it is easy to see why there were probs (contact evg@...) - hence another reason to stay 32bit.
4. Still in 32-bit, the app still needed some changes to accommodate the dialect 3 datatypes. Thus, once in FB1.5, we do not see an easy path back to the FB1.0/IB4/6 scenario - that has to be dropped.
5. -> 64-bit and FB2.x, I would suggest is another stage after having ensured FB1.5 is totally stable - shortcutting the stages is not an option. Every step must be safe and secure, robust and reliable.
Do read Evgeniy's paper. HTH
Kind Regards,Brian Gooch
[Non-text portions of this message have been removed]