Set autotrace on sample set of data of 410 million rows

Question

The dev env and the prod env is very different from our shop.
So if I do a sample on the prod data and then use the same query logic and do a real run and with autotrace on
Can I trust the plan and execution time of this sampling test and do an estimate on how it is gonna run in prod?
I guess I can use this to do a rough estimate of my prod run?
Any of you guys have done similar exercise?
I have the data size of 410 millions on my prod to process.
-Thx so much

Answer 1

I am talking about doing a perf exercise in the production environment but I am using only a sample cause and take 0.01 percent
idea of what I am doing. It is not linearly scale up of course but perhaps it gives me a rough idea?
insert into pipetest_result1
17:39:27   2  with pipetest1_smp001 as (
17:39:27   3  select * from pipetest1 sample(1) --can this give me a rough idea                                                  of how long it is gonna run?
17:39:27   4  ),
17:39:27   5   j1 as (
17:39:27   6  select id,value1,value2,pipetest1_YN(NVL(value1,1),NVL(value2,1))                                                  result from pipetest1_smp001
17:39:27   7  )
17:39:27   8  select id,result from j1 where j1.result='Y'
17:39:27   9
spool off
@pipetest1
set autotrace on
spool &1
Enter value for 1: pipetest_perf_feb15x1
insert into pipetest_result1
17:40:14   2  with pipetest1_smp001 as (
17:40:14   3  select * from pipetest1 sample(1) --can this give me a rough idea                                                  of how long it is gonna run?
17:40:14   4  ),
17:40:14   5   j1 as (
17:40:14   6  select id,value1,value2,pipetest1_YN(NVL(value1,1),NVL(value2,1))                                                  result from pipetest1_smp001
17:40:14   7  )
17:40:14   8  select id,result from j1 where j1.result='Y';
1231 rows created.
Elapsed: 00:00:02.13
Execution Plan
| Id  | Operation           | Name      | Rows  | Bytes | Cost (%CPU)|
|   0 | INSERT STATEMENT    |           |     1 |    51 |   399   (2)|
|   1 |  TABLE ACCESS SAMPLE| PIPETEST1 |     1 |    51 |   399   (2)|
Note
   - 'PLAN_TABLE' is old version
Statistics
        129  recursive calls
         47  db block gets
       2343  consistent gets
       2211  physical reads
      23536  redo size
        819  bytes sent via SQL*Net to client
       1003  bytes received via SQL*Net from client
          3  SQL*Net roundtrips to/from client
          2  sorts (memory)
          0  sorts (disk)
       1231  rows processed
spool off
@"pipetest1.sql"
set autotrace on
spool &1
Enter value for 1: pipetest1_feb15_all
insert into pipetest_result1
17:40:56   2  with pipetest1_smp001 as (
17:40:56   3  select * from pipetest1 --sample(1) --can this give me a rough ide                                                 a of how long it is gonna run?
17:40:56   4  ),
17:40:56   5   j1 as (
17:40:56   6  select id,value1,value2,pipetest1_YN(NVL(value1,1),NVL(value2,1))                                                  result from pipetest1_smp001
17:40:56   7  )
17:40:56   8  select id,result from j1 where j1.result='Y';
123924 rows created.
Elapsed: 00:00:47.60
Execution Plan
| Id  | Operation         | Name      | Rows  | Bytes | Cost (%CPU)|
|   0 | INSERT STATEMENT  |           | 11180 |   425K|   900  (57)|
|   1 |  TABLE ACCESS FULL| PIPETEST1 | 11180 |   425K|   900  (57)|
Note
   - 'PLAN_TABLE' is old version
Statistics
        719  recursive calls
       2901  db block gets
       3227  consistent gets
          0  physical reads
    2261104  redo size
        825  bytes sent via SQL*Net to client
       1005  bytes received via SQL*Net from client
          3  SQL*Net roundtrips to/from client
          1  sorts (memory)
          0  sorts (disk)
     123924  rows processed