Still uncertain whether to backup your database with RMAN or not?
RMAN significantly improved database backup by giving the following advantages:
- Faster backup.
- Internal management of backup and restore which are independent of OS or PLSQL scripts.
- Single interface for all OS.
- Internal Oracle capabilities such as binaries shrinking algorithm, parallelism in backup and restore procedures.
But here is another major benefit that might help you make the right decision:
BMR – Block Media Recovery
Follow this link to the original article posted in Hebrew:
http://www.oracle.com/global/il/technology/ido_ben_zeev.html
Installing the management agent (formerly known as intelligent agent) is usually a simple task as long as you asked your system administrator to open all the required ports.
When it comes to a windows cluster it involves a little bit more caution and precision. The agent works on a cluster environment no matter if you are using Oracle FailSafe or not. The following pdf simplifies the procedure:
Install agent on windows cluster
db_file_multiblock_read_count parameter helps us controlling the numbers of blocks being brought up to the buffer cache in one I/O action. In older oracle releases, this parameter was usually set between 4 and 16 on OLTP environments and up to 64 on DWH environment.
Starting 10gR2, db_file_multiblock_read_count is being set automatically depends on your OS I/O size and buffer cache size.
What happens is that even when you tune your system by optimizing the multiblock parameter, when you then upgrade to 11g (Fresh installation) Oracle might set up the default to an irrelevant number. So, for instance, if you work on RAC OLTP based system, and your multiblock is set on 8 you should expect a change of this parameter with major performance degradation.
I recently visited a customer who suffered from some nasty waits on “gc buffer busy acquire” as well as other GC related wait events right after upgrading to 11g. Further investigation and a stress test validated the assumption that this is due to the change of the multiblock parameter (from 16 to 128).
The reason for this behavior is clear- since one node keeps a large amount of blocks in the buffer cache, and the second node tries to acquire access to the same blocks (intense OLTP operations) it will wait on cache fusion events.
So be conscious about it!
Ido
Oracle introduced the asmcd utility already in 10g database, but with limited shell commands. Although we expected oracle to come up with advanced shell capabilities in 11g , a little has been done, but still very helpful.
A very nice feature allows us to migrate (export) asm files out to an external file system. The big plus here is that no database needs to be mounted on the machine where we would like to mount the asm disk groups on. For instance, you can use a storage utility to create a snapshot of your ASM disks and present them on a different machine where you want to them to be backed up. This is nice especially where you have a backup tool (evault , for example) that does not support ASM.
In the attached pdf, you will find all the steps and scripts to design such a configuration.
To make it easier on everyone, I put everything down in the following pdf:
Using ASM CP command
Principal Database Consultant
