The SQL TRACE facility and TKPROF are two basic performance diagnostic tools which can help you monitor and tune applications that run against Oracle. For best results, use these tools together rather than using EXPLAIN PLAN alone.

The SQL trace facility provides performance information on individual SQL statements. It generates the following statistics for each statement:

You can enable the SQL trace facility for a session or for an instance. When the SQL trace facility is enabled, performance statistics for all SQL statements executed in a user session or in an instance are placed into a trace file. The additional overhead of running the SQL trace facility against an application with performance problems is normally insignificant, compared with the inherent overhead caused by the application's inefficiency.

How to Use the SQL TRACE Facility and TKPROF
Follow these steps to use the SQL trace facility and TKPROF.

   1. Set initialization parameters for trace file management.
        ALTER SESSION SET TIMED_STATISTICS = TRUE;
or
       ALTER SYSTEM SET TIMED_STATISTICS = TRUE;

You can also select the name of the trace file with:
    alter session set tracefile_identifier = "My_trace_file";

   2. Enable the SQL trace facility for the desired session and run your application. This step produces a trace file containing statistics for the SQL statements issued by the application.
        ALTER SESSION SET SQL_TRACE = TRUE;                  to start trace
        ALTER SESSION SET SQL_TRACE = FALSE;               to stop trace
or
    SQL>EXECUTE dbms_session.set_sql_trace (TRUE);             to start trace
    SQL>EXECUTE dbms_session.set_sql_trace (FALSE);         to stop trace
or
    SQL>EXECUTE dbms_support.start_trace;                        to start trace
    SQL>EXECUTE dbms_support.stop_trace;                         to stop trace

If you get ORA-01031: insufficient privileges, then you need to get privileges:
GRANT ALTER SESSION TO xxx;

To Trace another session:
First you need to give privileges to a user to use this package:
    CREATE PUBLIC SYNONYM DBMS_SUPPORT FOR SYS.DBMS_SUPPORT;
    GRANT EXECUTE ON DBMS_SYSTEM to <user/role>;
     or
    GRANT EXECUTE ON DBMS_SYSTEM to PUBLIC;

Using the TOP Unix command or the v$session view you can check the SID and SERIAL# number of the session to monitor.
select sid, serial#, substr(terminal,1,10) terminal, substr(program,1,40) program
from v$session
where terminal in('OCR04','IMGSVR')
-- where username = 'FRAUDGUARD';;

Then, you activate the trace. There are several ways to do this, including the following three.

Run the files dbmssupp.sql and prvtsupp.plb from $ORACLE_HOME/rdbms/admin as SYS and grant execute on DBMS_SUPPORT to system;
    CREATE PUBLIC SYNONYM DBMS_SUPPORT FOR SYS.DBMS_SUPPORT;
    GRANT EXECUTE ON DBMS_SYSTEM to <user/role>;
     or
    GRANT EXECUTE ON DBMS_SYSTEM to PUBLIC;

TURN ON:
    exec dbms_support.start_trace_in_session( SID, SERIAL, waits=> true, binds=> true)
/* code to be traced executes during this time window */
TURN OFF:
    exec dbms_support.stop_trace_in_session( SID, SERIAL)

2- Using DBMS_MONITOR (10g only)
This package solves many of the complicated diagnostic data collection problems resulting from connection pooling and multithreading. Instead of specifying an Oracle Database session to be traced, you will be able to specify a service, module, or action to trace:

Example using SID and SERIAL
TURN ON: exec dbms_monitor.session_trace_enable(SID, SERIAL, waits=> true, binds=> true)
/* code to be traced executes during this time window */
TURN OFF: exec exec dbms_monitor.session_trace_disable(SID, SERIAL)

Example using Service or Module:
BEGIN
-->Enable/Disable Client Identifier Trace.
DBMS_MONITOR.client_id_trace_enable (client_id => ‘my_id’);
DBMS_MONITOR.client_id_trace_disable (client_id => ‘my_id’);

-->Enable/Disable Service, Module and Action Trace.
DBMS_MONITOR.serv_mod_act_trace_enable (service_name=>'my_service');
DBMS_MONITOR.serv_mod_act_trace_enable (service_name=>'my_service', module_name=>'my_module');
DBMS_MONITOR.serv_mod_act_trace_enable (service_name=>'my_service',module_name=>‘my_module’,action_name=>'INSERT');
DBMS_MONITOR.serv_mod_act_trace_disable (service_name=>'my_service',module_name=>'my_module',action_name =>'INSERT');

-->Enable/Disable Session Trace .
DBMS_MONITOR.session_trace_enable;
DBMS_MONITOR.session_trace_enable(session_id => 12, serial_num => 1011);
DBMS_MONITOR.session_trace_disable(session_id => 12, serial_num => 1011);
END;
/
With the DBMS_MONITOR package, Oracle offers a fully supported means of activating and deactivating diagnostic data collection for a specific business action (or actions) you want to trace.

3- Using SET_SQL_TRACE_IN_SESSION (but not recommended because doesn't catch extended SQL Tracing)
Do not use the DBMS_SUPPORT procedure called SET_SQL_TRACE_IN_SESSIONt, his procedure does not allow you to specify that you want waits and binds data in your trace file.

TURN ON: exec sys.dbms_system.set_sql_trace_in_session(SID, SERIAL#, true);
/* code to be traced executes during this time window */
TURN OFF: exec sys.dbms_system.set_sql_trace_in_session(SID, SERIAL#, false);

Using Trace Analyzer
Trace Analyzer is available only from Oracle Metalink (Note 224270.1 ).
Is a collection of sql files that generates an HTML Report from a trace file on the server. Its installation is very simple:
- Unzip THIS file to you local computer.
- Go to the Install directory and execute as SYS the file tacreate.sql. It will create the user TRCANLZR and will prompt for a password, tablespace location and connection string to the DB
- If you want to uninstall this, connect as SYS and run tadrop.sql
- Then when you want to analyze a trace file just perform the following:
sqlplus TRCANLZR/passwd@conn_string @full_path\run\trcanlzr.sql trace_file_on_the_server
After that, an html report will be generated from the place that you executed the previous command.

Tabular Statistics in TKPROF
TKPROF lists the statistics for a SQL statement returned by the SQL trace facility in rows and columns. Each row corresponds to one of three steps of SQL statement processing. Statistics are identified by the value of the CALL column:

COUNT: Number of times a statement was parsed, executed, or fetched. Generally high numbers for anything is cause of investigation (simply because nothing is 'free' in Oracle), but in particular, high numbers of parse calls could be a performance killer in multi-user systems.
CPU: Total CPU time in seconds for each process. This value is zero (0) if TIMED_STATISTICS is not turned on.
ELAPSED: The elapsed time represents the time (in seconds) that was taken to identify and modify rows during statement execution on the server. This is not the same as the time taken to return that information back to the client
DISK: Total number of data blocks physically read (or physical I/O) from the datafiles on disk for all parse, execute, or fetch calls. Generally you want to see blocks being read from the buffer cache rather than disk.
QUERY: Total number of buffers retrieved in consistent mode for all parse, execute, or fetch calls. Usually, buffers are retrieved in consistent mode for queries. A Consistent mode buffer is one that has been generated to give a consistent read snapshot for a long running transaction. The buffer actually contains this status in its header.
CURRENT: Total number of buffers retrieved in current mode. Buffers are retrieved in current mode for statements such as INSERT, UPDATE, and DELETE. This column is incremented if a buffer found in the buffer cache that is new enough for the current transaction and is in current mode (and it is not a CR buffer). This applies to buffers that have been read in to the cache as well as buffers that already exist in the cache in current mode.
ROWS: Total number of rows processed by the SQL statement. This total does not include rows processed by subqueries of the SQL statement.

For SELECT statements, the number of rows returned appears for the fetch step. For UPDATE, DELETE, and INSERT statements, the number of rows processed appears for the execute step.

First I think we need to make sure that these are hard and not soft parses. "Hard" parses are real parses so we want to minimize those. "Soft" parses occur when we are simply looking for that statement and are not as costly. You can use tkprof to determine if there is problem with hard parsing:

Look in the TKPROF output for the lines displayed after the "Total" PARSE, EXECUTE, FETCH block. In particluar, we are looking for the following lines:
"Misses in library cache during parse: n"
"Misses in Library Cache during Execute: n"

Misses in library cache during parse: 0 -> indicates that the statement was found in the library cache (soft parse).

The true parsing is listed in 'Misses in library cache during parse'
The "Misses in library cache during parse: 1" indicates only 1 hard parse. The "Parse" value in TKPROF includes both hard and soft parses. Soft parse does not involve parsing the statement, it just hashes the statement and compares this with the hashed value in the SQL area.
The count of parses to executions should be as close to one as possible. If there are a high number of parses per execution, this could indicate the lack of use of bind variables in SQL statements or poor cursor reuse.

Hard Parse:
Statement not previously executed, needs to be parsed, syntax check, semantic check, permissions check, allocate space from shared pool. Requires a library cache latch AND a shared pool latch.

Soft Parse:
Statement previously parsed. Syntax and check is not required, if the user is different than the previously parsing user then partial semantic check is still required, as object names may resolve to different objects. Requires only library cache latch.

Understanding Recursive Calls
Sometimes, in order to execute a SQL statement issued by a user, Oracle must issue additional statements. Such statements are called recursive calls or recursive SQL statements. For example, if you insert a row into a table that does not have enough space to hold that row, then Oracle makes recursive calls to allocate the space dynamically. Recursive calls are also generated when data dictionary information is not available in the data dictionary cache and must be retrieved from disk.
If recursive calls occur while the SQL trace facility is enabled, then TKPROF produces statistics for the recursive SQL statements and marks them clearly as recursive SQL statements in the output file. You can suppress the listing of Oracle internal recursive calls (for example, space management) in the output file by setting the SYS command-line parameter to NO. The statistics for a recursive SQL statement are included in the listing for that statement, not in the listing for the SQL statement that caused the recursive call. So, when you are calculating the total resources required to process a SQL statement, consider the statistics for that statement as well as those for recursive calls caused by that statement.

Library Cache Misses in TKPROF
TKPROF also lists the number of library cache misses resulting from parse and execute steps for each SQL statement. These statistics appear on separate lines following the tabular statistics. If the statement resulted in no library cache misses, then TKPROF does not list the statistic. In "Sample TKPROF Output", the statement resulted in one library cache miss for the parse step and no misses for the execute step.

There are a huge number of sort options that can be accessed by simply typing 'TkProf' at the command prompt. A useful starting point is the 'fchela' sort option which orders the output by elapsed time fetching (remember that timing information is only available with timed_statistics set to true in the "init.ora" file). The resultant .prf file will contain the most time consuming SQL statement at the start of the file. Another useful parameter is sys. This can be used to prevent SQL statements run as user SYS from being displayed. This can make the output file much shorter an easier to manage.

EXPLAIN = Uses the EXPLAIN PLAN statement to generate the execution plan of each SQL statement traced. TKPROF also displays the number of rows processed by each step of the execution plan
SYS = With the value of NO causes TKPROF to omit recursive SQL statements from the output file. In this way you can ignore internal Oracle statements such as temporary table operations.
SORT = Causes TKPROF to sort the SQL statements in order of the sum of the CPU time spent executing and the CPU time spent fetching rows before writing them to the output file. For greatest efficiency, always use SORT parameters

Deciding Which Statements to Tune
You need to find which SQL statements use the most CPU or disk resource.
If the TIMED_STATISTICS parameter is on, then you can find high CPU activity in the CPU column. If TIMED_STATISTICS is not on, then check the QUERY and CURRENT columns.
With the exception of locking problems and inefficient PL/SQL loops, neither the CPU time nor the elapsed time is necessary to find problem statements. The key is the number of block visits, both query (that is, subject to read consistency) and current (that is, not subject to read consistency). Segment headers and blocks that are going to be updated are acquired in current mode, but all query and subquery processing requests the data in query mode. These are precisely the same measures as the instance statistics CONSISTENT GETS and DB BLOCK GETS.
You can find high disk activity in the disk column.

Every TKPROF report starts with a header that lists the TKPROF version, the date and time the report was generated, the name of the trace file, the sort option used, and a brief definition of the column headings in the report. Every report ends with a series of summary statistics. You can see the heading and summary statistics on the sample TKPROF report shown earlier in this paper.

The main body of the TKPROF report consists of one entry for each distinct SQL statement that was executed by the database server while SQL trace was enabled. There are a few subtleties at play in the previous sentence. If an application queries the customers table 50 times, each time specifying a different customer_id as a literal, then there will be 50 separate entries in the TKPROF report. If however, the application specifies the customer_id as a bind variable, then there will be only one entry in the report with an indication that the statement was executed 50 times. Furthermore, the report will also include SQL statements initiated by the database server itself in order to perform so-called “recursive operations” such as manage the data dictionary and dictionary cache.

The entries for each SQL statement in the TKPROF report are separated by a row of asterisks. The first part of each entry lists the SQL statement and statistics pertaining to the parsing, execution, and fetching of the SQL statement. Consider the following example:

Reading across, we see that while SQL trace was enabled, the application called on the database server to parse this statement once. 0.01 CPU seconds over a period of 0.02 elapsed seconds were used on the parse call, although no physical disk I/Os or even any buffer gets were required. (We can infer that all dictionary data required to parse the statement were already in the dictionary cache in the SGA.)

The next line shows that the application called on Oracle to execute the query once, with less than 0.01 seconds of CPU time and elapsed time being used on the execute call. Again, no physical disk I/Os or buffer gets were required. The fact that almost no resources were used on the execute call might seem strange, but it makes perfect sense when you consider that Oracle defers all work on most SELECT statements until the first row is fetched.

The next line indicates that the application performed 14 fetch calls, retrieving a total of 194 rows. The 14 calls used a total of 0.59 CPU seconds and 0.99 seconds of elapsed time. Although no physical disk I/Os were performed, 33,633 buffers were gotten in consistent mode (consistent gets). In other words, there were 33,633 hits in the buffer cache and no misses. I ran this query from SQL*Plus, and we can see here that SQL*Plus uses an array interface to fetch multiple rows on one fetch call. We can also see that, although no disk I/Os were necessary, it took quite a bit of processing to complete this query.

The remaining lines on the first part of the entry for this SQL statement show that there was a miss in the library cache (the SQL statement was not already in the shared pool), the CHOOSE optimizer goal was used to develop the execution plan, and the parsing was performed in the RSCHRAG schema.

Notice the text in square brackets concerning recursive depth. This did not actually appear on the report—I added it for effect. The fact that the report did not mention recursive depth for this statement indicates that it was executed at the top level. In other words, the application issued this statement directly to the database server. When recursion is involved, the TKPROF report will indicate the depth of the recursion next to the parsing user.

There are two primary ways in which recursion occurs. Data dictionary operations can cause recursive SQL operations. When a query references a schema object that is missing from the dictionary cache, a recursive query is executed in order to fetch the object definition into the dictionary cache. For example, a query from a view whose definition is not in the dictionary cache will cause a recursive query against view$ to be parsed in the SYS schema. Also, dynamic space allocations in dictionary-managed tablespaces will cause recursive updates against uet$ and fet$ in the SYS schema.

Use of database triggers and stored procedures can also cause recursion. Suppose an application inserts a row into a table that has a database trigger. When the trigger fires, its statements run at a recursion depth of one. If the trigger invokes a stored procedure, the recursion depth could increase to two. This could continue through any number of levels.

So far we have been looking at the top part of the SQL statement entry in the TKPROF report. The remainder of the entry consists of a row source operation list and optionally an execution plan display. (If the explain keyword was not used when the TKPROF report was generated, then the execution plan display will be omitted.) Consider the following example, which is the rest of the entry shown above:

The row source operation listing looks very much like an execution plan. It is based on data collected from the SQL trace file and can be thought of as a “poor man’s execution plan”. It is close, but not complete.

The execution plan shows the same basic information you could get from the autotrace facility of SQL*Plus or by querying the plan table after an EXPLAIN PLAN statement—with one key difference. The rows column along the left side of the execution plan contains a count of how many rows of data Oracle processed at each step during the execution of the statement. This is not an estimate from the optimizer, but rather actual counts based on the contents of the SQL trace file.

Although the query in this example goes against a dictionary view and is not terribly interesting, you can see that Oracle did a lot of work to get the 194 rows in the result: 11,146 range scans were performed against the i_obj2 index, followed by 11,146 accesses on the obj$ table. This led to 12,665 non-unique lookups on the i_obj# index, 11,339 accesses on the tab$ table, and so on.

In situations where it is feasible to actually execute the SQL statement you wish to explain (as opposed to merely parsing it as with the EXPLAIN PLAN statement), I believe TKPROF offers the best execution plan display. GUI tools such as TOAD will give you results with much less effort, but the display you get from TOAD is not 100% complete and in certain situations critical information is missing. (Again, my experience is with the free version!) Meanwhile, simple plan table query scripts like my explain.sql presented earlier in this paper or utlxpls.sql display very incomplete information. TKPROF gives the most relevant detail, and the actual row counts on each operation can be very useful in diagnosing performance problems. Autotrace in SQL*Plus gives you most of the information and is easy to use, so I give it a close second place.

Another Example:
The following listing shows TKPROF output for one SQL statement as it appears in the output file:

SELECT *
FROM emp, dept
WHERE emp.deptno = dept.deptno;

call   count      cpu    elapsed     disk    query current    rows
---- ------- ------- --------- -------- -------- ------- ------
Parse     11     0.08      0.18        0       0       0         0
Execute   11     0.23      0.66        0       3       6         0
Fetch     35     6.70      6.83      100   12326       2       824
------------------------------------------------------------------
total     57     7.01      7.67      100   12329       8       826
Misses in library cache during parse: 0

If it is acceptable to have 7.01 CPU seconds and to retrieve 824 rows, then you need not look any further at this trace output. In fact, a major use of TKPROF reports in a tuning exercise is to eliminate processes from the detailed tuning phase.

You can also see that 10 unnecessary parse call were made (because there were 11 parse calls for this one statement) and that array fetch operations were performed. You know this because more rows were fetched than there were fetches performed.

The information displayed in a TKPROF report can be extremely valuable in the application tuning process. Of course the execution plan listing will give you insights into how Oracle executes the SQL statements that make up the application, and ways to potentially improve performance. However, the other elements of the TKPROF report can be helpful as well.

Looking at the repetition of SQL statements and the library cache miss statistics, you can determine if the application is making appropriate use of Oracle’s shared SQL facility. Are bind variables being used, or is every query a unique statement that must be parsed from scratch?

From the counts of parse, execute, and fetch calls, you can see if applications are making appropriate use of Oracle’s APIs. Is the application fetching rows one at a time? Is the application reparsing the same cursor thousands of times instead of holding it open and avoiding subsequent parses? Is the application submitting large numbers of simple SQL statements instead of bulking them into PL/SQL blocks or perhaps using array binds?

Looking at the CPU and I/O statistics, you can see which statements consume the most system resources. Could some statements be tuned so as to be less CPU intensive or less I/O intensive? Would shaving just a few buffer gets off of a statement’s execution plan have a big impact because the statement gets executed so frequently?

The row counts on the individual operations in an execution plan display can help identify inefficiencies. Are tables being joined in the wrong order, causing large numbers of rows to be joined and eliminated only at the very end? Are large numbers of duplicate rows being fed into sorts for uniqueness when perhaps the duplicates could have been weeded out earlier on?

Step 1 - Look at the totals at the end of the tkprof output

OVERALL TOTALS FOR ALL NON-RECURSIVE STATEMENTS

call   count      cpu    elapsed     disk    query current    rows
---- ------- ------- --------- -------- -------- ------- ------
Parse   A 7     1.87      4.53      385 G 553      22         0
Execute E 7     0.03      0.11    P   0   C   0    D 0     F 0
Fetch   E 6     1.39      4.21    P 182 C 820    D 3     F 20
------------------------------------------------------------------
Misses in library cache during parse: 5
Misses in library cache during execute: 1

      8 user SQL statements in session.
     12 internal SQL statements in session.
[B] 54 SQL statements in session.
      3 statements EXPLAINed in this session.

1. Compare [A] & [B] to spot over parsing. In this case we have 7 parses for 54 statements which is ok.

2. You can use [P], [C] & [D] to determine the hit ratio.

Hit Ratio is logical reads/physical reads:
Logical Reads = Consistent Gets + DB Block Gets
Logical Reads = query + current
Logical Reads = Sum[C] + Sum[D]
Logical Reads = 0+820 + 0+3
Logical Reads = 820 + 3
Logical Reads = 823

Hit Ratio = 1 - (Physical Reads / Logical Reads)
Hit Ratio = 1 - (Sum[P] / Logical Reads)
Hit Ratio = 1 - (128 / 823)
Hit Ratio = 1 - (0.16)
Hit Ratio = 0.84 or 84%

3. We want fetches to be less than the number of rows as this will mean we have done less work (array fetching). To see this we can compare [E] and [F].
[E] = 6 = Number of Fetches
[F] = 20 = Number of Rows
So we are doing 6 fetches to retrieve 20 rows - not too bad.
If arrayfetching was configured then rows could be retrieved with less fetches. Remember that an extra fetch will be done at the end to check that the end of fetch has been reached.

4. [G] Shows reads on the Dictionary cache for the statements.
In this case we have done 553 reads from the Library cache.

call   count      cpu    elapsed     disk    query current    rows
---- ------- ------- --------- -------- -------- ------- ------
Parse      1        7       122        0 0       0         0
Execute    1       75       461 5 H 297    I 3     J 1
Fetch      0        0         0    0 0       0     0
------------------------------------------------------------------

[H] shows that this query is visiting 297 blocks to find the rows to update
[I] shows that only 3 blocks are visited performing the update
[J] shows that only 1 row is updated.

call   count      cpu    elapsed    disk   query current    rows
---- ------- ------- --------- ------- -------- ------- ------
Parse   M 2 N 221       329      0    45       0       0
Execute O 3 P    9      17    0   0    0 0
Fetch      3        6         8    0 L 4    0 K 1
------------------------------------------------------------------
Misses in library cache during parse: 2 [Q]

[K] is shows that the query has returned 1 row.
[L] shows that we had to read 4 blocks to get this row back.
This is fine.

[M] show that we are parsing the statement twice - this is not desirable especially as the cpu usage is high [N] in comparison to the execute figures : [O] & [P]. [Q] shows that these parses are hard parses. If [Q] was 1 then the statement would have had 1 hard parse followed by a soft parse (which just looks up the already parsed detail in the library cache).

This is not a particularly bad example since the query has only been executed a few times. However excessive parsing should be avoided as far as possible by:

o Ensuring that code is shared:
- use bind variables
- make shared pool large enough to hold query definitions in memory long enough to be reused.

To make a quick assessment if it may be poorly tuned SQL causing the problem, the following query will help:

select sid, name, value
from v$sesstat ss, v$statname sn
where ss.statistic# = sn.statistic#
and name in ('db block gets', 'consistent gets','physical reads','table fetch contiued row')
and sid = &SID
order by 3 desc;

Also, it is important to identify whether or not the session is waiting or actually doing work. The following query will help asses this.

select substr(event, 1, 30) "Resource", time_waited / 100 "Time waited/used (in seconds)"
from v$session_event
where sid = &SID
union
select substr(name, 1, 30) "Resource", value / 100 "Time waited/used (in seconds)"
from v$statname sn, v$sesstat ss
where sn.statistic# = ss.statistic#
and sn.name = 'CPU used by this session'
and ss.sid = &SID
order by 2 desc ;

If you experience a lot of waits, it may be of some help to query v$session_wait to find the parameters for the wait events. However, querying v$session_wait takes a while relative to the length of most waits. Hope this helps

Tracing with Triggers
In some cases, you may want to trace when an user connects to the database and only stop when he disconnects. On such ocassions, we can make use of database event level triggers to enable/disable that automatically.

create or replace trigger trace_trigger_scott_on
after logon on database
when (user = 'SCOTT')
declare
   stmt varchar2(100);
   hname varchar2(20);
   uname varchar2(20);
begin
   select sys_context('USERENV','HOST'), sys_context('USERENV','SESSION_USER') into hname, uname
      from dual;
   stmt := 'alter session set tracefile_identifier =' || hname|| '_' || uname;
   execute immediate stmt;
   execute immediate 'alter session set sql_trace=TRUE';
end;
/

create or replace trigger trace_trigger_scott_off
before logoff on database
when (user = 'SCOTT')
begin
execute immediate 'alter session set sql_trace=false';
end;
/

Application Tuning Using Extended SQL Trace And TKPROF

Introduction

Sometimes it’s useful to be able to trace an Oracle session. This could be because you’ve written a batch process, it’s taking a long time to run, and you want to find out what’s causing it to run slow. It could also be because you’ve written a web or desktop application, and users complain that certain parts of the application run too slow. By tracing an Oracle session, which contains one or more SQL statements executed in order, you want find out precisely what statements your session executed, how they were executed by Oracle, and what delays occurred as the statements were executed.

This document sets out how to enable SQL tracing for any Oracle database (8i, 9i, 10g), how to format and interpret the trace files, and where to look next to resolve your performance issues.

Scenario 1 : PL/SQL Simulate Workload, single process.

To help explain how tracing in Oracle works, I have provided some PL/SQL code that simulates two typical workloads, with code at the start of the procedure that turns on extended SQL tracing for the rest of the session. The code works off of the sample data within the SCOTT/TIGER schema found within every Oracle database. The first bit of PL/SQL code simulates a typical workload.

CREATE OR REPLACE procedure simulate_workload_with_binds

var1 varchar2(100);

begin

-- turn on extended SQL tracing

execute immediate 'alter session set timed_statistics = true';

execute immediate 'alter session set max_dump_file_size = unlimited';

execute immediate 'alter session set tracefile_identifier = ''test1_binds_single_process''';

execute immediate 'alter session set events ''10046 trace name context forever, level 8''';

-- main body of code

for c in 1 .. 500 loop

insert into test (col1) values (c);

commit;

end loop;

commit;

for c in 1 .. 50 loop

update test

set col1 = col1;

commit;

end loop;

for c in 1 .. 500 loop

begin

select col1 into var1

from test

where col1 = c;

exception

when others then null;

end;

end loop;

for c in 1 .. 500 loop

insert into test(col1) values (c);

commit;

end loop;

commit;

for c in 1 .. 500 loop

delete from test where col1=c;

commit;

end loop;

commit;

execute immediate 'alter session set events ''10046 trace name context off''';

end;

points to note :

The following code turns on what’s termed ‘extended SQL tracing’ for this session:

-- turn on extended SQL tracing

execute immediate 'alter session set timed_statistics = true';

execute immediate 'alter session set max_dump_file_size = unlimited';

execute immediate 'alter session set tracefile_identifier = ''test_1 _binds_single_process''';

execute immediate 'alter session set events ''10046 trace name context forever, level 8''';

“alter session set timed_statistics=true” ensures that the trace data contains timings for the various events and statements contained within it.

“alter session set max_dump_file_size = unlimited” ensures that your trace doesn’t prematurely end because you’ve reached the default max dump file size.

“alter session set tracefile_identifier = ‘’test1_binds_single_process’’” puts the specified text into the trace file name, making it easier to identify later

“alter session set events ‘’10046 trace name context forever, level 8’’’” is the important line, and tells Oracle to emit trace data, including information on ‘wait events’ (this is what the level 8 bit refers to) to the trace file.

Note the bit about ‘extended SQL tracing’. It’s called ‘extended’ because it’s like regular SQL tracing (which you can turn on by issuing the command Alter session set sql_trace=true ) but it additionally captures diagnostic information on what’s called ‘wait events’ – behind the scenes events that delay your SQL statement – events such as locks on tables, contention for disk blocks and so on. Also, with later versions of Oracle, you can active extended SQL trace using new built in Oracle packages - however, the method detailed above works for all Oracle versions, doesn’t require you to install any additional packages, and captures all the information we need for tracing. Note also that it’s possible to interactively turn on tracing for a session using the SQL*Plus prompt, and you can also turn it on for sessions other than your own – however for simplicity’s sake I’m embedding the relevant code directly into the PL/SQL procedure. For more details on the various ways to invoke extended SQL tracing see here

The PL/SQL code is the compiled, and then executed with the commands :

sQL> exec simulate_workload;

PL/SQL procedure successfully completed.

The next step is to use TKPROF (provided by default with all recent Oracle installations) to process and report on this raw trace file. From the DOS command line, find the raw trace file, and issue the command:

TKPROF raw_trace_file_name formatted_trace_file_name

e.g.

tkprof markr9i_ora_5820_test1_binds_single_process.trc markr9i_ora_5820_test1_binds_single_process.txt

Then, open up the formatted trace file (download here) with notepad, textpad or similar.

The formatted trace file has the following sections:

The preamble that tells you about TKPROF formatted trace files, and the SQL statement that enabled extended SQL trace.

TKPROF: Release 9.2.0.5.0 - Production on Mon Oct 4 14:19:34 2004

Trace file: markr9i_ora_5820_test1_binds_single_process.trc

Sort options: default

********************************************************************************

count = number of times OCI procedure was executed

cpu = cpu time in seconds executing

elapsed = elapsed time in seconds executing

disk = number of physical reads of buffers from disk

query = number of buffers gotten for consistent read

current = number of buffers gotten in current mode (usually for update)

rows = number of rows processed by the fetch or execute call

********************************************************************************

alter session set events '10046 trace name context forever, level 8'

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 0 0.00 0.00 0 0 0 0

Execute 1 0.00 0.00 0 0 0 0

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 1 0.00 0.00 0 0 0 0

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

********************************************************************************

The statements that were executed as part of the PL/SQL packages.

INSERT INTO TEST (COL1)

VALUES

(:B1)

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 500 0.09 0.08 0 1 1007 500

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 501 0.09 0.08 0 1 1007 500

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

********************************************************************************

COMMIT

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 7 0.00 0.00 0 0 0 0

Execute 1553 0.08 0.07 0 0 1550 0

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 1560 0.08 0.07 0 0 1550 0

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

********************************************************************************

UPDATE TEST SET COL1 = COL1

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 50 6.12 6.19 0 452 27078 25000

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 51 6.12 6.19 0 452 27078 25000

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

Elapsed times include waiting on following events:

Event waited on Times Max. Wait Total Waited

---------------------------------------- Waited ---------- ------------

undo segment extension 293 0.00 0.00

********************************************************************************

select file#

from

file$ where ts#=:1

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 1 0.00 0.00 0 0 0 0

Fetch 2 0.00 0.01 2 3 0 1

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 4 0.00 0.01 2 3 0 1

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: SYS (recursive depth: 2)

Rows Row Source Operation

------- ---------------------------------------------------

1 TABLE ACCESS BY INDEX ROWID FILE$

1 INDEX RANGE SCAN I_FILE2 (object id 42)

Elapsed times include waiting on following events:

Event waited on Times Max. Wait Total Waited

---------------------------------------- Waited ---------- ------------

db file sequential read 2 0.00 0.01

********************************************************************************

SELECT COL1

FROM TEST WHERE COL1 = :B1

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 500 0.01 0.00 0 0 0 0

Fetch 500 0.09 0.08 0 4500 0 500

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 1001 0.10 0.09 0 4500 0 500

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

********************************************************************************

INSERT INTO TEST(COL1)

VALUES (:B1)

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 500 0.09 0.09 0 2 1026 500

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 501 0.09 0.09 0 2 1026 500

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

********************************************************************************

DELETE FROM TEST

WHERE COL1=:B1

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 500 0.25 0.25 0 4500 1516 1000

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 501 0.25 0.25 0 4500 1516 1000

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

********************************************************************************

alter session set events '10046 trace name context off'

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 1 0.00 0.00 0 0 0 0

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 2 0.00 0.00 0 0 0 0

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

Taking the first PL/SQL commands and looking at them more detail

for c in 1 .. 500 loop

insert into test (col1) values (c);

commit;

end loop;

This results in the following formatted trace output:

INSERT INTO TEST (COL1)

VALUES (:B1)

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 500 0.09 0.08 0 1 1007 500

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 501 0.09 0.08 0 1 1007 500

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

********************************************************************************

The PL/SQL commands set up a loop that executes 500 times, with the value of the loop counter being inserted into the ‘test’ table each time the loop executes.
What the trace file tells us is that a single command was parsed once, then executed 500 times. This is because PL/SQL uses bind variables by default, and Oracle would therefore only need to hard parse the statement once, with subsequent executions being able to use the stored parsed value of the statement each time a new value was inserted. Therefore, this section of the trace file refers to 500 separate SQL code executions, and the trace file has wrapped these up into a single section of the trace file.
It’s also useful to look at the ‘totals’ at the bottom of the statement breakdown, that show you the cpu time taken up by the statements and the total elapsed time. Note that the elapsed time is shorter than the CPU time – this is a rounding error and infact the elapsed time in this case should be considered to be the CPU time or a few microseconds more.

Once you have reviewed the rest of the SQL statements in the trace file, move on to the final part of the trace file.

The trace file summary

OVERALL TOTALS FOR ALL NON-RECURSIVE STATEMENTS

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 0 0.00 0.00 0 0 0 0

Execute 0 0.00 0.00 0 0 0 0

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 0 0.00 0.00 0 0 0 0

Misses in library cache during parse: 0

OVERALL TOTALS FOR ALL RECURSIVE STATEMENTS

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 14 0.00 0.00 0 0 0 0

Execute 3606 6.64 6.70 0 4955 32177 27000

Fetch 502 0.09 0.09 2 4503 0 501

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 4122 6.73 6.80 2 9458 32177 27501

Misses in library cache during parse: 0

Elapsed times include waiting on following events:

Event waited on Times Max. Wait Total Waited

---------------------------------------- Waited ---------- ------------

undo segment extension 293 0.00 0.00

db file sequential read 2 0.00 0.01

14 user SQL statements in session.

1 internal SQL statements in session.

15 SQL statements in session.

********************************************************************************

Trace file: markr9i_ora_5820_test1_binds_single_process.trc

Trace file compatibility: 9.02.00

Sort options: default

1 session in tracefile.

14 user SQL statements in trace file.

1 internal SQL statements in trace file.

15 SQL statements in trace file.

9 unique SQL statements in trace file.

6053 lines in trace file. 5728 lines in trace file.

The main Area of interest in this section is the ‘Overall Totals For All Recursive Statements’ section. This is the SQL statements contained within your PL/SQL package and comprises the totals for all parses, executes and fetches incurred by your statements. In this particular bit of PL/SQL, bind variables are used and statements are executed many times, which leads to a much smaller amount of parses than executes – which is generally a good thing.

If your parses were more or less the same as your executes, you should check your code to make sure bind variables are being used – in PL/SQL, it is difficult to not use bind variables, but with Java, VB etc you have to consciously use them, and therefore it’s not unusual to find excessive amounts of parsing in their traced code, which is generally a bad thing as parsing ties up the CPU and prevents the application scaling if many users are using the program concurrently.

Note also the totals, where the elapsed time is only slightly higher than the CPU time – this shows that the program only took slightly longer to execute than the CPU time it required, meaning that it wasn’t unduly delayed by waiting for contended resources, network traffic and so on. If you were tuning this application, you wouldn’t really worry about this sort of profile.

Lastly, although it’s not really an issue for this simulated workload, note also the columns headed ‘query’ ‘current’ and ‘rows’ – these detail the amount of rows, blocks and so on that the query requested and worked on – if your program appears to be requesting and processing an excessive number of rows, look to reduce these (by altering the SQL statements) to bring down the total amount of data the query is looking to process.

Scenario 2 : PL/SQL Simulated Workload, 5 concurrent processes.

This scenario looks at a more common situation, where a program that previously performed without too much of an issue now is exponentially slower when a larger number of copies are run concurrently. To test this, a line in the PL/SQL program was changed to alter the trace file identifier:

-- turn on extended SQL tracing

execute immediate 'alter session set timed_statistics = true';

execute immediate 'alter session set max_dump_file_size = unlimited';

execute immediate 'alter session set tracefile_identifier = ''test2_binds_concurrent_processes''';

execute immediate 'alter session set events ''10046 trace name context forever, level 8''';

and then five copies of the program were executed simultaneously, by opening up five DOS command windows and then executing the PL/SQL program within each session at the same time.

The first thing that you will notice is that each copy of the program takes considerably longer to run when run concurrently with the others. In short, this is because each copy is trying to insert, update, select from and delete from the same table, leading to contention for resources.

You will notice that five separate trace files have been output into the /udump directory, with a different process ID embedded within each filename. Assuming that all processes were started at the same time, take one trace file (download here) and process it using TKPROF:

tkprof markr9i_ora_2652_test2_binds_concurrent_processes.trc markr9i_ora_2652_test2_binds_concurrent_processes.txt

Then, open up the formatted trace file (download here) and view using Notepad.exe. You will find the same structure within the trace file, the same order of statements and so on. However, looking at for example the set of SQL statements analyzed before, note this time that the elapsed time is twice as much as the CPU time.

INSERT INTO TEST (COL1)

VALUES (:B1)

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 500 0.14 0.31 0 2 1021 500

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 501 0.14 0.31 0 2 1021 500

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

********************************************************************************

Also, look at the part of the trace file that analyzes the SQL statement

UPDATE TEST SET COL1 = COL1

In our original trace file (where the command was run in isolation) the trace file shows the following analysis:

UPDATE TEST SET COL1 = COL1

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 50 6.13 6.27 0 451 25600 25000

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 51 6.13 6.27 0 451 25600 25000

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

********************************************************************************

whilst in our current trace file, where the command was run concurrently, the trace file analysis shows:

UPDATE TEST SET COL1 = COL1

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 50 114.77 781.11 6 277901 128304 124500

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 51 114.77 781.11 6 277901 128304 124500

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

Elapsed times include waiting on following events:

Event waited on Times Max. Wait Total Waited

---------------------------------------- Waited ---------- ------------

enqueue 305 3.12 605.91

buffer busy waits 146 0.08 1.38

undo segment extension 162 0.00 0.00

db file sequential read 6 0.06 0.12

********************************************************************************

Our original trace file shows a total elapsed time of 6.27 seconds and a CPU time of 6.13 seconds, which is acceptable. However our current trace file shows a total elapsed time of 781.11 seconds and a CPU time of 114.77 seconds – our CPU time has risen dramatically, and our elapsed time is several times longer again.

However, note the section that follows the statement breakdown, headed by ‘Elapsed Times Include Waiting On The Following Events’. This is called the wait event analysis, and it tells us exactly what kept our statement waiting.

The key column is the ‘total waited’ column, and you can see in this instance that our statement was kept waiting for 605.91 seconds by the ‘enqueue’ wait event – which means ‘locks’. Looking back at our PL/SQL code, this problem is happening because each of the five copies of the program are trying to update the same columns in the same table, and more often than not the row is locked when a copy of the application needs to write to it.

Also, note the ‘buffer busy waits’ event. Buffer busy waits (http://www.dba-oracle.com/art_builder_bbw.htm) occur when an Oracle session needs to access a block in the buffer cache, but cannot because the buffer copy of the data block is locked. This buffer busy wait condition can happen for either of the following reasons:

The block is being read into the buffer by another session, so the waiting session must wait for the block read to complete.
Another session has the buffer block locked in a mode that is incompatible with the waiting session's request.

Buffer busy waits are happening in our system because each session is trying to access the same blocks in the database buffer, meaning that most of them have to wait until the other one has finished reading the buffer block. However, note that whilst the number of buffer busy waits is 50% of the enqueue waits, the total waited time is only a fraction of the wait time caused by enqueue events – therefore, if you were tuning this application, you would focus the vast majority of your time on solving the enqueue problem, as this will have the greatest impact on your application (i.e. it will potentially remove 605.91 seconds from the time your session takes to run, rather than the 1.38 seconds that would be removed by dealing with the buffer busy waits issue)

The rest of our statements in the formatted trace file also suffer from the buffer busy waits problem, but the total delay is very small compared to our enqueue problem detailed before.

SELECT COL1

FROM TEST WHERE COL1 = :B1

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 500 0.00 0.00 0 0 0 0

Execute 500 0.00 0.00 0 0 0 0

Fetch 500 0.17 1.08 0 2499 0 500

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 1500 0.17 1.09 0 2499 0 500

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

Rows Row Source Operation

------- ---------------------------------------------------

2 TABLE ACCESS FULL TEST

Elapsed times include waiting on following events:

Event waited on Times Max. Wait Total Waited

---------------------------------------- Waited ---------- ------------

buffer busy waits 27 0.05 0.23

latch free 2 0.01 0.01

********************************************************************************

INSERT INTO TEST(COL1)

VALUES (:B1)

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 500 0.09 0.56 0 149 1010 500

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 501 0.09 0.56 0 149 1010 500

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

Elapsed times include waiting on following events:

Event waited on Times Max. Wait Total Waited

---------------------------------------- Waited ---------- ------------

buffer busy waits 4 0.05 0.13

********************************************************************************

DELETE FROM TEST

WHERE COL1=:B1

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 1 0.00 0.00 0 0 0 0

Execute 500 0.71 3.91 0 8466 845 566

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 501 0.71 3.91 0 8466 845 566

Misses in library cache during parse: 0

Optimizer goal: CHOOSE

Parsing user id: 59 (recursive depth: 1)

Elapsed times include waiting on following events:

Event waited on Times Max. Wait Total Waited

---------------------------------------- Waited ---------- ------------

buffer busy waits 81 0.06 1.11

latch free 11 0.03 0.06

enqueue 3 0.09 0.16

********************************************************************************

Note that the last statement that deletes from the test table also has issues with locking and latches, but the total contribution to total session time is very small.

Next, look at the summary section of the formatted trace file.

********************************************************************************

OVERALL TOTALS FOR ALL NON-RECURSIVE STATEMENTS

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 0 0.00 0.00 0 0 0 0

Execute 0 0.00 0.00 0 0 0 0

Fetch 0 0.00 0.00 0 0 0 0

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 0 0.00 0.00 0 0 0 0

Misses in library cache during parse: 0

OVERALL TOTALS FOR ALL RECURSIVE STATEMENTS

call count cpu elapsed disk query current rows

------- ------ -------- ---------- ---------- ---------- ---------- ----------

Parse 512 0.00 0.00 0 0 0 0

Execute 3605 115.77 787.28 6 286518 132493 126066

Fetch 500 0.17 1.08 0 2499 0 500

------- ------ -------- ---------- ---------- ---------- ---------- ----------

total 4617 115.94 788.37 6 289017 132493 126566

Misses in library cache during parse: 0

Elapsed times include waiting on following events:

Event waited on Times Max. Wait Total Waited

---------------------------------------- Waited ---------- ------------

enqueue 308 3.12 606.08

buffer busy waits 258 0.08 2.85

undo segment extension 162 0.00 0.00

db file sequential read 6 0.06 0.12

latch free 15 0.03 0.10

513 user SQL statements in session.

0 internal SQL statements in session.

513 SQL statements in session.

********************************************************************************

Trace file: markr9i_ora_2652_test2_binds_concurrent_processes.trc

Trace file compatibility: 9.02.00

Sort options: default

1 session in tracefile.

513 user SQL statements in trace file.

0 internal SQL statements in trace file.

513 SQL statements in trace file.

8 unique SQL statements in trace file.

9491 lines in trace file.

Looking at the ‘Overall Totals For All Recursive Statements’ section, it shows that the total CPU time was 115.94 and the total elapsed time was 788.37, with the difference between the two being the amount of time Oracle was having to wait around for resources or other wait events.
The ‘Elapsed times include waiting on following events’ section then summarises all of the wait events, with ‘enqueue’ again being by far the largest contributor to the total time taken up by the session.

The outcome of tracing this session would be that you would concentrate your efforts on resolving the locking problems being experienced by the UPDATE statement in your code, and this would have a major effect on the total running time of your application.

(Footnote : We did in fact also have a wait event section in the previous formatted trace file markr9i_ora_5820_test1_binds_single_process.txt:

Elapsed times include waiting on following events:

Event waited on Times Max. Wait Total Waited

---------------------------------------- Waited ---------- ------------

undo segment extension 293 0.00 0.00

db file sequential read 2 0.00 0.01

However, the total waiting time was virtually nil – 0.01 seconds - and would not be something you would need to concern yourself with when tuning a particular session.)

Conclusions

Tuning your batches, statements and applications using extended SQL trace and TKPROF allows you to examine exactly what statements your code is executing, whether they used bind variables, how many rows they processes and what wait events occurred whilst they were trying to execute. By tuning your applications using this approach, you can determine exactly what is slowing your code down, allowing you to concentrate your efforts on eliminating those waits that add the most to your total execution time.

Tuning applications in this way takes much of the ‘guess work’ out of tuning Oracle applications and makes it a process based on logical steps and documented approaches rather than picking solutions out of the air and hoping they’l have an effect.

Note : Extended SQL Tracing When Using Connection Pooling

One problem that comes up with extended SQL tracing is when your application is an application (typically Java, or .NET) that uses connection pooling. In this sort of situation, the statements for a typical user session can end up spread over several separate trace files, each one relating to a separate pooled connection. In addition (and this is the worst bit) each trace file can contain SQL statements for other users who are using the same pooled connection, mixing up their statements with yours.

Connection pooling is dealt with in the above books and articles, but in summary the problem is usually dealt with in one of the following ways.

Oracle 10g has specific features (detailed in Tracing SQL in Oracle Database 10g by Kimberly Floss) which allow you to identify just those SQL statements that pertain to a particular user/application.
Force all users to log off the system then run the application, and cut and paste the separate trace files into one before analyzing using TKPROF
Run a separate copy of the application on a different port number (for example, port 81) and have this version connect using a dedicated connection, and then TKPROF the resultant trace file as normal.

Introduction

Scenario 1 : PL/SQL Simulate Workload, single process.

Conclusions

Further Reading

Note : Extended SQL Tracing When Using Connection Pooling