Hi Nenad!
Oracle is not SQL Server. Oracle tenant databases are not as well
isolated as SQL Server databases. Technically, Oracle multi-tenant
solution is very much inferior to SQL Server, MySQL or PostgreSQL in
many aspects. Backup and copying the databases is one such aspect. Also,
SQL Server can have some tenants in simple transaction mode which
corresponds to Orracle's NOARCHIVELOG while having other tenants in the
full transaction mode, which corresponds to archive logging. However,
the existence of the root container and CDB_* tables is sometimes very
useful. And I agree with you: I would like to see tenants that I can
backup and restore to another instance. I would also like to be able to
create indexes on the local temporary tables.
On the other side, I would also like seeing PL/SQL implemented on MS
SQL. If PostgreSQL can have a version of PL/SQL and DB2 can have it, I
am not quite sure why MS is so hesitant with that. MS SQL Server can,
for instance, execute Python from within the database:
https://www.sqlshack.com/how-to-use-python-in-sql-server-2017-to-obtain-advanced-data-analytics/
So, there is obviously a capability to replace T-SQL with something
else. I believe this is a marketing decision. And that "machine
learning" stuff is a story by itself. Looks like MS is on track to
produce Heuristic Algorithmic Language computer, HAL 9000. The BSOD will
be replaced by "You're killing me Dave".
On 12/23/20 1:04 PM, Noveljic Nenad wrote:
--
That works - apart that it wipes out the whole destination CDB. It’s a strange behavior for the feature that was marketed as a consolidation panacea. Also, there doesn’t seem to be a way to specify a new name for the cloned PDB. In contrast, duplicate from active leaves the CDB and all other PDBs intact. It’s difficult for me to understand the reason for the completely different behavior in both use cases. No such problems in MS SQL Server, by the way.
Thanks,