You are on page 1of 1

HOW TO: Analyzing and Interpreting AWR Report | DBA Group

1 de 1

http://dbagroup.cl/blog/?p=537

HOW TO: Analyzing and Interpreting AWR Report


Compartir este post:

Hello
Today I wanted to share with you my personal approach for how to get into the details on AWR Report and where to see
and where should be our attention. Keep in mind just two main focuses : the Application and the Database instance
1) Application: Most of the times you should go directly to review how was the SQL statements Performance. That is
MANDATORY and the result of your analysis will be the classic TOP 10 SQL with the worst performance. That should
be called : Oracle Applications Performance Tuning .
2) Instance: read once, twice or three times the AWR looking for issues. What kind of issues or metrics ?
For example here I have a Conclusion and Workaround for one database instance issue regarding performance :
1) Tuning recommendations for PROD-ACME database instance:
Based on the statistics of I/O per event within the database I recommended :
I saw that for a total of 22MB most of the times their waits are due to lack of buffers cache (used to put
there disk blocks with tables data) with the rows used. I found 86% of wait events are due this event for the
database instance.
Action Item: Need to increase of MB for buffers cache within the database in order to increase the available
blocks and space in memory to keep the most used rows there.
How much ? In order to see how much we need to increase my recommendation is to put
DB_CACHE_ADVICE initialization parameter to ON. Then review (after some weeks) and query the
V$DB_CACHE_ADVICE system view so there you will get what is the optimal point to set buffers cache
for this instance. (That Advisor was a new feature of 9i release) .
e.g. Waits and What Represents.
What does it mean ?
Function Name
Direct Reads
Buffer Cache
Reads
Direct Writes
Others
LGWR
DBWR
Streams AQ
TOTAL:

Reads:
Data
1.9T
533.8G

Reqs per
sec
54.28
313.70

Data per
sec
23.4386
6.32998

Writes:
Data
628M
0M

Reqs per
sec
0.06
0.00

Data per
sec
.007273
0M

Waits:
Count
0
19M

7.3G
17.9G
6M
29M
3M
2.5T

0.12
4.78
0.01
0.02
0.00
372.90

.086315
.212123
.000069
.000335
.000034
30.0675

283.4G
45.1G
24.7G
12.1G
0M
365.9G

15.42
3.30
68.79
8.12
0.00
95.68

3.36128
.534663
.292555
.143538
0M
4.33931

0
445.2K
2966.9K
1878
292
22.4M

Avg
Tm(ms)
2.79

2.05
2.00
3.69
33.73
2.67

So you can see above that WAITS amount of Buffer Cache Reads represents the 85% of Waits for this instance. In
this case this was one of the bottlenecks.
2) Tuning recommendations for ERP ACME Suite and Customs Applications :
Lets see next section of the same example as follows :
Based on the awrrpt_1_12809_12883.html that I got here in my desk I reviews this section to see top 5 events of waits :
Top 5 Timed Foreground Events
Event
Waits
Time(s) Avg wait (ms) % DB time Wait Class
db file sequential read 53,182,945 124,071 2
30.56
User I/O
12/10/2015 0:49

You might also like