I'm trying to reduce the time to install Oracle on a pipeline we have. At the moment, it's installed using dbca in silent mode with a database template file. The majority of the time seems to be creating the data dictionary etc. I have a theory that it would be quicker to clone a totally vanilla install and add the application specific data using data pump after. I just wanted to see if it's worth the effort trying this - has anyone tried it?
FYI - it's Oracle 12c EE using ASM for storage.
You can try and compare this yourself with DBCA.
When you create a new database with DBCA and choose the Custom Database option, that does not include datafiles, and creates a database from scratch by running scripts to build the dictionary and other objects. This may take 15-60 minutes depending on the hardware and options selected.
When you choose the General Purpose option, that in fact performs an RMAN restore of a database that was already prepared and included in the database software. This should not take more than 5 minutes. It is indeed faster. (I do not like it though, because it contains extra cost options such as Spatial or OLAP, and I prefer to keep the database small and simple, so by default I skip all options, even Java and Text).
With DBCA, you can use an existing database as source for creating a template that includes datafiles as well, so yes, you can build your initial database, create a template from it, and reuse it.
