Phycial Reads / Logical Reads best course of action?

From: Thiko! <biwombi_at_hotmail.com>
Date: 23 May 2005 05:16:58 -0700
Message-ID: <1116850618.656860.98590@z14g2000cwz.googlegroups.com>

Using TOAD's session browser while performing a test in a 'before and after' scenario I would like to know if the following is a reasonable increase in traffic. There is an increase in the logical reads and in the consistent gets between Test 1 and Test 2. Would indexes on any large tables involved help? I have a brief understanding of what physical and logical reads are.

Fig 3. Toad Block Gets Consistent Gets*** Physical Reads Logical Reads**

TEST1

3.0.84 caching_pass1	69			20189			1858			20630
3.0.84 caching_pass2	401 (470)		5869 (26058)		1859 (3717)		5898 (26528)

TEST2
3.0.85 caching_pass1	441			20189			1856			20630
3.0.85 caching_pass2	49 (490)		11479 (31668)		1859 (3715)	        11528

(32158)

• When a block is requested by a query, Oracle looks for it in the Buffer Cache and if found, it results in a logical read and if it does not find the Block in there it results in a physical read (disk I/O).

The two tests also show there to be a 6 fold increase in the number of selects taking place in Test2.

Fig 2. EOAdaptor SELECT Statements Table1 Table2 Table3 Other

TEST1

3.0.84 caching_pass1	427			53	101	53	n
3.0.84 caching_pass2	189			0	5	0	n

TEST2
3.0.85 caching_pass1	426			53	101	53	n
3.0.85 caching_pass2	1266			314	450	314	n

Many thanks!

Thiko! Received on Mon May 23 2005 - 07:16:58 CDT