13 March, 2015

Parallel Execution -- 2b PX Servers

Continuing my previous post, here I demonstrate  using V$SQLSTATS.PX_SERVERS_EXECUTIONS and a couple of issues around it.

I have restarted the database.

[oracle@localhost ~]$ sqlplus hemant/hemant

SQL*Plus: Release 11.2.0.2.0 Production on Fri Mar 13 22:49:20 2015

Copyright (c) 1982, 2010, Oracle.  All rights reserved.


Connected to:
Oracle Database 11g Enterprise Edition Release 11.2.0.2.0 - Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options

HEMANT>show parameter cpu

NAME                                 TYPE        VALUE
------------------------------------ ----------- ------------------------------
cpu_count                            integer     4
parallel_threads_per_cpu             integer     4
resource_manager_cpu_allocation      integer     4
HEMANT>select degree from user_tables where table_name = 'LARGE_TABLE';

DEGREE
----------------------------------------
         1

HEMANT>select /*+ PARALLEL */ count(*) from Large_Table;

  COUNT(*)
----------
   4802944

HEMANT>select executions, px_servers_executions, sql_fulltext
  2  from v$sqlstats
  3  where sql_id = '8b0ybuspqu0mm';

EXECUTIONS PX_SERVERS_EXECUTIONS SQL_FULLTEXT
---------- --------------------- --------------------------------------------------------------------------------
         1                    16 select /*+ PARALLEL */ count(*) from Large_Table

HEMANT>
HEMANT>select /*+ PARALLEL */ count(*) from Large_Table;

  COUNT(*)
----------
   4802944

HEMANT>select executions, px_servers_executions, sql_fulltext
  2  from v$sqlstats
  3  where sql_id = '8b0ybuspqu0mm';

EXECUTIONS PX_SERVERS_EXECUTIONS SQL_FULLTEXT
---------- --------------------- --------------------------------------------------------------------------------
         2                    32 select /*+ PARALLEL */ count(*) from Large_Table

HEMANT>

[Note : To understand why the executions took 16 PX Servers inspite of the degree on table being 1, see this post]
So we see that PX_SERVERS_EXECUTIONS shows cumulative statistics.  Let's try a slight twist.

HEMANT>connect / as sysdba
Connected.
SYS>alter system set parallel_max_servers=8;

System altered.

SYS>connect hemant/hemant
Connected.
HEMANT>select /*+ PARALLEL */ count(*) from Large_Table;

  COUNT(*)
----------
   4802944

HEMANT>select executions, px_servers_executions, sql_fulltext
  2  from v$sqlstats
  3  where sql_id = '8b0ybuspqu0mm';

EXECUTIONS PX_SERVERS_EXECUTIONS SQL_FULLTEXT
---------- --------------------- --------------------------------------------------------------------------------
         3                    40 select /*+ PARALLEL */ count(*) from Large_Table

HEMANT>

Because I set PARALLEL_MAX_SERVERS to 8, my query on Large_Table could take only 8 PX Servers at the next execution.  V$SQLSTATS.PX_SERVERS_EXECUTIONS now shows a cumulative count of 40 for 3 executions. There is no way to determine how many PX Servers were used in each of the 3 executions, because the history of executions is not maintained.
(In my controlled experiment, we know, by deduction, that the 3rd execution took 8 PX Servers simply because we know already that the first 2 executions took a cumulative count of 32 PX Servers -- by deducting 32 from 40 to get 8 for the 3rd execution)

What happens if the SQL is invalidated ?

HEMANT>alter table large_table parallel 4;

Table altered.

HEMANT>select /*+ PARALLEL */ count(*) from Large_Table;

  COUNT(*)
----------
   4802944

HEMANT>select executions, px_servers_executions, sql_fulltext
  2  from v$sqlstats
  3  where sql_id = '8b0ybuspqu0mm';

EXECUTIONS PX_SERVERS_EXECUTIONS SQL_FULLTEXT
---------- --------------------- --------------------------------------------------------------------------------
         1                     8 select /*+ PARALLEL */ count(*) from Large_Table

HEMANT>

The ALTER TABLE, being a DDL, had invalidated the query on Large_Table.  So, V$SQLSTATS also got reset.  Therefore, EXECUTIONS reset to 1 and PX_SERVES_EXECUTIONS got reset to 8.

.
.

.

1 comment:

Sayan Malakshinov said...

It could be a little easier if we can use real time sql monitor(>=11g):

https://gist.github.com/xtender/c024cd87ff86cb213c94