Restore single datafile from source database to target database.

Question

Here's my issue:
Database Release : 11.2.0.3 across both the source and targets. (ARCHIVELOG mode)
O/S: RHEL 5 (Tikanga)
Database Storage: Using ASM on a stand-alone server (NOT RAC)
Using Oracle GG to replicate changes on the Source to the Targets.
My scenario:
We utilize sequences to keep the primary key in tact and these are replicated utilizing GG. All of my schema tables are located in one tablespace and datafile and all of my indexes are in seperate tablespace (nothing is being partitioned).
In the event of media failure on the Target or my target schema being completely out of whack, is there a method where I can copy the datafile/tablespace from my source (which is intact) to my target?
I know there are possibilites of
1) restore/recover the tablespace to a SCN or timestamp in the past and then I could use GoldenGate to run the transactions in (but this could take time depending on how far back I need to recover the tablespace and how many transactions have processed with GG) (This is not fool-proof).
2) Could use DataPump to move the data from the Source schema to the Target schema (but the sequences are usually out of order if they haven't fired on the source, get that 'sequence is defined for this session message'). I've tried this scenario.
3) I could alter the sequences to get them to proper number using the start and increment by feature (again this could take time depending on how many sequences are out of order).
I would think you could
1) back up the datafile/tablespace on the source,
2)then copy the datafile to the target.
3) startup mount;
4) Newname the new file copied from the source (this is ASM)
5) Restore the datafile/tablespace
6) Recover the datafile/tablespace
7) alter database open;
Question 1: Do I need to also copy the backup piece from the source when I execute the backup tablespace on the source as indicated in my step 1?
Question 2: Do I need to include "plus archivelog" when I execute the backup tablespace on the source as indicated in my step 1?
Question 3: Do I need to execute an 'alter system switch logfile' on the Target when the recover in step 6 is completed?
My scenario sounds like a Cold Backup but running with Archivelog mode, so the source could be online while the database is running.
Just looking for alternate methods of recovery.
Thanks,
Jason

Answer 1

Let me take another stab at sticking a fork into this myth about separating tables and indexes.
Let's assume you have a production Oracle database environment with multiple users making multiple requests and the exact same time. This assumption mirrors reality everywhere except in a classroom where a student is running a simple demo.
Let's further assume that the system looks anything like a real Oracle database system where the operating system has caching, the SAN has caching, and the blocks you are trying to read are split between memory and disk.
Now you want to do some simple piece of work and assume there is an index on the ename column...
SELECT * FROM emp WHERE ename = 'KING';The myth is that Oracle is going to, in parallel, read the index and read the table segments better, faster, whatever, if they are in separate physical files mapped by separate logical tablespaces somehow to separate physical spindles.
Apply some synapses to this myth and it falls apart.
You issue your SQL statement and Oracle does what? It looks for those index blocks where? In memory. If it finds them it never goes to disk. If it does not it goes to disk.
While all this is happening the hundreds or thousands of other users on the database are also making requests. Oracle is not going to stop doing work while it tries to find your index blocks.
Now it finds the index block and decides to use the ROWID value to read the block containing the row with KING's data. Did it freeze the system? Did it lock out everyone else while it did this? Of course not. It puts your read request into the queue and, again, first checks memory to see if it needs to go to disk.
Where in here is there anything that indicates an advantage to having separate physical files?
And even if there was some theoretical reason why separate files might be better ... are they separate in the SAN's cache? No. Are they definitely located on separate stripes or separate physical disks? Of course not.
Oracle uses logical mappings (tables and tablespaces) and SANS use logical mappings so you, the DBA or developer, have no clue as to where anything physically is located.
PS: Ouija Boards don't work either.