TUNING LOG BUFFER

Hi all,

        I look for your valuable feedback for the following scenario.         

        Oracle Version is 7.1.6         

        OS Version is Solaris 2.5.1         

        In our Database the v$sysstat view shows a value more than zero for the parameter named 'redo log space requests'. This value is growing regularly, till
we stop and start the database.         

        The recommendation is to increase the LOG_BUFFER to a higher value. Our current value is 1MB. (Our db_block size is 2K and the default value is 4 times
of db_block size)

        I have a strong feeling that for this size of log buffer we should not get redo log space requests and relog space request wait time, since the
database is generating 20 to 30 MB of redolog information everyday and 1000 to
1700 transactions between log switches only.         

        To support this, I have noticed 2 entries of log buffer, one as 16000 and another as 1024000 in sgastat view of this database. But every other database has got only one entry for log buffer with value as we specify in init
file.         

SGA Info. :-         

NAME                      VALUE
-------------------- ----------
Fixed Size                48260
Variable Size          22321952
Database Buffers       10240000
Redo Buffers            1024000
	

SYSSTAT Info. :-

STATISTIC# NAME                                                    VALUE
---------- -------------------------------------------------- ----------
        83 redo log space requests                                    22

(Database started on 30/08/97).

SGASTAT Info. :-

NAME                            BYTES
-------------------------- ----------
log_buffer                      16000
log_buffer                    1024000

Is there anybody, who got an explanation for this, or a method to look further into this.         

All your suggestions are most welcome.         

Thanks in advance

P.Manivannan Received on Tue Sep 02 1997 - 00:00:00 CDT