This content has been marked as final. Show 4 replies
The BACKUP_TAPE_IO_SLAVES parameter is used to keep tape continuously moving by simulating asynchronous tape I/O.
Transfer rate can also be seen at Oracle level with V$BACKUP_SYNC_IO and V$BACKUP_ASYNC_IO.
I don't believe there is any other used of this parameter.
SET lines 200 col filename FOR a40 SET pages 50 SELECT a.device_type, a.TYPE, s.status, s.mbytes_processed AS "Total MB", a.filename, ROUND(a.bytes/(1024*1024)) AS "Size MB", TO_CHAR(a.open_time,'dd-mon-yyyy hh24:mi:ss') AS open_time, ROUND(a.elapsed_time/100) AS "Time (s)", ROUND(a.elapsed_time/100) AS "Elapsed (s)", ROUND(a.effective_bytes_per_second/(1024*1024)) AS "MB/s" FROM v$backup_async_io a, v$rman_status s WHERE a.rman_status_recid = s.recid ORDER BY a.use_count;
You can try looking at these to get memory information :
As far as memory goes the simple thing is to use ASMM so Oracle will automatically size and adjust memory for you.
select * from v$shared_pool_advice; select * from v$sga_target_advice;
You might want to keep the following in mind:
The SGA large pool is only used if you set DBWR_IO_SLAVES for disk and/or BACKUP_TAPE_IO_SLAVES for tape buffering. If LARGE_POOL_SIZE is not set it will get memory from the shared pool. If the server cannot get enough memory, it will allocate buffers from PGA and use synchronous I/O and you will see a message in the alert log.
Thanks for reply.
I am using ASSM.
Right now I am taking RMAN backup on tape as well as on Disk also..
I have checked both the views that is V$BACKUP_SYNC_IO and V$BACKUP_ASYNC_IO.
In V$BACKUP_SYNC_IO it is showing TAPE backup info and in V$BACKUP_ASYNC_IO it is showing Disk backup info..
Parameter backup_tape_io_slaves set to false and dbwr_io_slaves value is 0..
So what is advisable for my system?
Should I set parameter backup_tape_io_slaves to true?
What is difference between ASYNC and SYNC?
Asynchronous is advantageous because a server process can perform more than 1 Input/Output operation at a time. So you want to use V$BACKUP_ASYNC_IO since Red Hat supports asynchronous.
On backup_tape_io_slaves you can test it on your system and see if your times improve for tape. They either do or don't, EFFECTIVE_BYTES_PER_SECONDS (V$BACKUP_ASYNC_IO) should be at a rate close to the capacity of your backup device.
Since you are already using ASSM you don't have to do much. You can try to tune this for hours and hours, but mostly likely its a waste of time.
Worth noting V$BACKUP_ASYNC_IO only holds information since your last instance restart. Same with the other but you are asynchronous so there should be no need to use it.