Can I leverage Db replication to help keep downtime to a minimum during system migration? ...looking for a sanity check here.
I’m in the process of migrating a company’s legacy enterprise system to new hardware and current version(s) of IDS and 4gl.
The company wants to relocate the new server to a facility in another state before it goes live.
The servers currently reside in the same facility.
Once the new server is in place/sync, the old server will be taken off-line.
(I know there are other steps here but I’m looking for a general sanity check)
I’m hoping for a little more sophistication rather than just shipping/transmitting backup data to the new server and reloading .
The environment is the classic IDS/4gl legacy set up running on an IBM/AIX platform.
The current production server is running Aix 5.3, IDS 10, 4gl 7.32
The new server is running Aix 6.1, IDS 11.7, 4gl 7.50.
True, you must run at least 11.70, but the migration path from 10 is direct. Most important is to check the data integrity and to indentify eventual ALTER IN PLACE statements here (and fix them, no big deal). the rest of the operation is very quick (a few minutes, according to the number of databases of your instance).
When your migration scenario has been well prepared and rehearsed, the process can take 15minutes or less of downtime, according to the number of databases your have in this instance. I have done this several times this year and last year on big instances.
So the first migration step from 10 to 11.70 is not such a big issue. I would still consider it.
When your situation has been well ready and practiced, the procedure can just could have taken 15minutes or less of recovery time, according to the variety of data source you have in this example. I have done this several periods this season and last season with a friend at a scale.