Data Buffer Quality Low !

[nicholas_chang]

Hi All,

During my system check on one of our client R3 Production system, i found out the data buffer quality is around 60% in ST04. What should i do to increase the percentage? Do i need to do database reorg ??

Your two cents of advise is kindly appreciate.

Thanks.

Rgds,

Nicholas.

[Former Member]

Hi,

Data buffer size is determined by the product of the block size (Oracle parameter DB_BLOCK_SIZE) and the number of database block buffers (Oracle parameter DB_BLOCK_BUFFERS) specified in the init<sid>.ora oracle parameter file. By default SAP uses a value of 8192 bytes for the DB_BLOCK_SIZE parameter for most Oracle databases. A database that has been created by whatever value used for this DB_BLOCK_SIZE parameter cannot be later changed for the life of the database. You can change the value for DB_BLOCK_BUFFERS.

In sdn Search for "best practices for performance tuning sap r/3 and oracle database configurations by Mr.Kostas G Gavrielidis...Its a very good document

Regards

Umesh K

[Former Member]

Increasing the size of the DB_BLOCK_BUFFERS is a simple <i>patch</i> that does not address the real problem.

By itself, to have the data buffer quality low means nothing. You have to analyze the reasons why you have it low.

And this requires an "indepth" analysis.

I'd suggest you to start analyzing your shared pool, the problem could be caused by expensive SQL statemens. Addressing them can improve your situation.

[nicholas_chang]

Hi Fidel,

How can i analyze the shared pool? Still, the quality of sharepool is good, around 90%. My priority is to enhance the DATA BUFFER quality, any ways to do it ? Will Reorganization the Database using brtools solve the problem?

Thanks.

[Former Member]

Hi Nicholas,

as Fidel already mentioned, you have to analyze your databases workload to find your problem.

May be one single statement ruins your buffer quality. A buffer quality of 60% is very bad, I see this in our systems only when, for some reason, the statistics get not updated. I would recommend first to run an update statistic with "<i>brconnect -c -u / -f stats -t ALL</i>" which at least should not have a negative impact.

If the quality stays at 60% you may have one or more statements doing full table scans. Check the number of full table scans in ST04 transactions. The number of rows fetched by rowid should be much higher than those by scan. I have no system available yet and don't remember the exact names of the fields, but you'll find what I mean. Another possibility is an invalid or missing index. Check in DB02 missing tables and indexes.

A reorganization only would reduce the table size if you have free space in the tables, but it would not solve the problem, as long as you don't have the index problem described above.

Best regards

Ralph