RE: RMAN - Some basic Qs.

From: Gaja Krishna Vaidyanatha <oraperfman_at_yahoo.com>
Date: Thu, 03 Apr 2003 09:28:53 -0800
Message-ID: <F001.0057989C.20030403092853@fatcity.com>

Vivek,

Answers to your questions embedded. You may want to read up on the manuals for more details. BTW, you don't have to copy my e-mail id in your replies, given that I belong to the list, just as many others. I am trying to avoid getting 2 copies of every e-mail that you send.

Cheers,

Gaja

• VIVEK_SHARMA <VIVEK_SHARMA_at_infosys.com> wrote:
  > Gaja ,
  >
  > Good indeed to see you back on the List .
  >
  > My Qs. in CAPITALS below :-
  >
  >
  > -----Original Message-----
  > Sent: Wednesday, April 02, 2003 2:14 AM
  > To: Multiple recipients of list ORACLE-L
  >
  >
  > All,
  >
  > The primary reason why RMAN does not generate
  > excessive redo is because because it does not put
  > the
  > tablespace in "hot backup mode". Thus any blocks
  > belonging to a given tablespace that are modified
  > before the END BACKUP command is processed, do not
  > require block-level before and after images. Hence
  > the
  > reduction in redo generation.
  >
  > So how does RMAN backup without "hot backups"?
  >
  > RMAN is aware of the format of an Oracle datafile,
  > and
  > reads datafiles for the backup in DB_BLOCK_SIZE
  > chunks. This is not the case with most OS utilities
  > (tar, cpio, cp, dd etc.), which read files in
  > 512-byte
  > OS blocksize chunks.
  >
  > IS BACKUP DATA AS AT POINT-OF-TIME OF START OF RMAN
  > BACKUP ?
  > IF SO , HOW IS DATA INTEGRITY ACROSS DATAFILES
  > MAINTAINED ?
  > HOW ARE BLOCK DATA VALUES AS AT POINT OF BACKUP
  > MAINTAINED WITHOUT STOPPING DMLs TO IT ?
  >
  

Gaja>:

There is no concept of "backup data at point-of-time start of the RMAN backup". Backups as such do not require "read consistency". A recovered
tablespace/database requires "consistency". This consistency is achieved by recovering the datafile or the entire database (if relevant) to the point noted in the controlfile, using the relevant online and archived redo logs. Ofcourse things are different for "incomplete recovery scenarios".

The backup is done "whenever it is done". Again, the issue of consistency arises only when a recovery is done. In the case of hot backups or RMAN backups, the redo generation continues during the backup (as required) and the issue of "consistency" is dealt with when recovery is performed. With hot backups, Oracle plays it safe and generates more redo, to protect against "block-fracture" on the backed-up files. This is required due to the 512-byte read-chunk-size that OS utilities use, which is different than DB_BLOCK_SIZE on most databases. This not relevant in RMAN backups due to DB_BLOCK_SIZE read-chunk-size and the relevant synchronization methods that RMAN uses.

>
> As a result, the issue of a fractured block (for
> which
> block-level before/after images are taken) on the
> destination where the backup is done, does not come
> into play in RMAN.
>
> MY UNDERSTANDING OF HOT BACKUP :-
>
> ASSUME 1 TABLESPACE HAS 2 DATAFILES & DMLs HAPPENING
> ONLY TO FILE 2.
> AFTER SWITCHING TABLESPACE TO BEGIN BACKUP .
> ASSUMING SEQUENTIAL O.S. BACKUP OF DATAFILES , WHILE
> BACKUP OF THE 1ST FILE IS UNDERWAY ,
> BEFORE IMAGES OF ALL DMLs HAPPENING TO FILE 2 ARE
> BEING ARCHIVED .
> THEREAFTER FILE 2 IS BACKED UP .
> FINALLY TABLESPACE END BACKUP IS ISSUED.
>
> THUS USING THE BEFORE IMAGES OF BLOCKS OF FILE 2 ,
> THE TABLESPACE CAN BE BROUGHT TO
> DATA EXISTING AS AT POINT OF START OF HOT BACKUP.
>

Gaja: See previous explanation. Your understanding is not fully accurate, please read the manuals for more details. Basically, it is OK for the backup copy of File#1 to be "different" from File# 2. The issue of consistency is dealt with during recovery and Oracle does a great job of it, by dealing with the logs that need to be applied, on a per file basis.

> Hope that helps,
>
> Gaja
>

==stuff deleted==

---

---

Do you Yahoo!?
Yahoo! Tax Center - File online, calculators, forms, and more http://tax.yahoo.com

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.net
-- 
Author: Gaja Krishna Vaidyanatha
  INET: oraperfman_at_yahoo.com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

Received on Thu Apr 03 2003 - 11:28:53 CST