Oracle Data Buffer Contents Scripts

The Oracle v$bh view shows the contents of the data buffers as well as the number of blocks for each type of segment in the buffer. This view is primarily useful for indicating the amount of table and index caching in multiple data blocks. Combining the v$bh view with dba_objects and dba_segments provides a block-by-block listing of the data buffer contents and indicates how well the buffers are caching tables and indexes. This is very important in Oracle10g since the data buffer sizes can be altered dynamically.

There are several data dictionary tricks that can be used when writing a script for mapping data objects to RAM buffers:

§ Duplicate object names: When joining dba_objects to dba_segments, the name, type, and owner are all required to distinguish the object sufficiently.

§ Multiple blocksizes: To show objects in the separate instantiated buffers such as db_2k_cache_size , etc., the block size for the object must be displayed. This is achieved by computing the block size from dba_segments , dividing bytes by blocks.

§ Partitions: With a standard equi-join, every object partition joins to every segment partition for a particular object. Hence, the following qualification is required to handle partitions:

§ Clusters: Clusters present a challenge when joining the v$bh row with its corresponding database object. Instead of joining the bh.objd to object_id, it needs to be joined into data_object_id .

§ Multiple caches: There are situations where a particular block may be cached more than once in the buffer cache. This is a mystifying concept, but it is easily overcome by creating the following in-line view:

-- *************************************************

-- This script is free for non-commercial purposes

-- with no warranties. Use at your own risk.

-- To license this script for a commercial purpose,

-- contact info@rampant.cc

-- *************************************************

set pages 999

set lines 92

ttitle 'Contents of Data Buffers'

drop table t1;

create table t1 as

select

o.owner owner,

o.object_name object_name,

o.subobject_name subobject_name,

o.object_type object_type,

count(distinct file# || block#) num_blocks

from

dba_objects o,

v$bh bh

where

o.data_object_id = bh.objd

and

o.owner not in ('SYS','SYSTEM')

and

bh.status != 'free'

group by

o.owner,

o.object_name,

o.subobject_name,

o.object_type

order by

count(distinct file# || block#) desc

;

column c0 heading "Owner" format a12

column c1 heading "Object|Name" format a30

column c2 heading "Object|Type" format a8

column c3 heading "Number of|Blocks in|Buffer|Cache" format 99,999,999

column c4 heading "Percentage|of object|blocks in|Buffer" format 999

column c5 heading "Buffer|Pool" format a7

column c6 heading "Block|Size" format 99,999

select

t1.owner c0,

object_name c1,

case when object_type = 'TABLE PARTITION' then 'TAB PART'

when object_type = 'INDEX PARTITION' then 'IDX PART'

else object_type end c2,

sum(num_blocks) c3,

(sum(num_blocks)/greatest(sum(blocks), .001))*100 c4,

buffer_pool c5,

sum(bytes)/sum(blocks) c6

from

t1,

dba_segments s

where

s.segment_name = t1.object_name

and

s.owner = t1.owner

and

s.segment_type = t1.object_type

and

nvl(s.partition_name,'-') = nvl(t1.subobject_name,'-')

group by

t1.owner,

object_name,

object_type,

buffer_pool

having

sum(num_blocks) > 10

order by

sum(num_blocks) desc

;

A sample listing from this exciting report is shown below. The report lists the tables and indexes that reside inside the data buffer at the exact moment that the script was executed. This is important information for the Oracle professional who needs to know how many blocks for each object reside in the RAM buffer. To effectively manage the limited RAM resources, the Oracle DBA must be able to know the ramifications of decreasing the size of the data buffer caches.

The following is the report from buf_blocks.sql when run against a large Oracle data warehouse.

This is an interesting report because there are three object types: tables; indexes; and partitions. The subsets of the DEFAULT pool for KEEP and RECYCLE are also evident. Also, all indexes are defined in the largest supported block size (db_32k_cache_size), and multiple buffer pools of 4K, 8K, 16K and 32K sizes are defined.

The output of this script can be somewhat confusing due to the repeated DEFAULT buffer pool name. In earlier releases of Oracle, the KEEP and RECYCLE buffer pools are subsets of db_cache_size and can ONLY accommodate objects with the DEFAULT db_block_size. In later releases, the KEEP and RECYCLE pools become independent pools.

Conversely, any block sizes that are NOT the default db_block_size, go into the buffer pool named DEFAULT. The output listing shows that there are really six mutually exclusive and independently sized buffer pools, four of them are called DEFAULT.