Offset and fetch in oracle sql developer - sql

Question

We have data in millions (total rows 1698393). While exporting this data in text takes 4 hours. I need to know if there is a way to reduce the exporting time for those many records from Oracle database using SQL Developer.
with cte as (
select *
from (
select distinct
system_serial_number,
( select s.system_status
from eim_pr_system s
where .system_serial_number=a.system_serial_number
) system_status,
( select SN.cmat_customer_id
from EIM.eim_pr_ib_latest SN
where SN.role_id=19
and SN.system_serial_number=a.system_serial_number
) SN_cmat_customer_id,
( select EC.cmat_customer_id
from EIM.eim_pr_ib_latest EC
where EC.role_id=1
and a.system_serial_number=EC.system_serial_number
) EC_cmat_customer_id
from EIM.eim_pr_ib_latest a
where a.role_id in (1,19)
)
where nvl(SN_cmat_customer_id,0)!=nvl(EC_cmat_customer_id,0)
)
select system_serial_number,
system_status,
SN_CMAT_Customer_ID,
EC_CMAT_Customer_ID,
C.Customer_Name SN_Customer_Name,
D.Customer_Name EC_Customer_Name
from cte,
eim.eim_party c,
eim.eim_party D
where c.CMAT_Customer_ID=SN_cmat_customer_id
and D.CMAT_Customer_ID=EC_cmat_customer_id;
offset first 5001 rows fetch next 200000 rows only

Answer 1

You can get rid of a lot of the joins and correlated sub-queries (which will speed things up by reducing the number of table scans) by doing something like:
SELECT a.system_serial_number,
s.system_status,
a.SN_cmat_customer_id,
a.EC_cmat_customer_id,
a.SN_customer_name,
a.EC_customer_name
FROM (
SELECT l.system_serial_number,
MAX( CASE l.role_id WHEN 19 THEN l.cmat_customer_id END ) AS SN_cmat_customer_id,
MAX( CASE l.role_id WHEN 1 THEN l.cmat_customer_id END ) AS EC_cmat_customer_id
MAX( CASE l.role_id WHEN 19 THEN p.customer_name END ) AS SN_customer_name,
MAX( CASE l.role_id WHEN 1 THEN p.customer_name END ) AS EC_customer_name
FROM EIM.eim_pr_ib_latest l
INNER JOIN
EIM.eim_aprty p
ON ( p.CMAT_Customer_ID= l.cmat_customer_id )
WHERE l.role_id IN ( 1, 19 )
GROUP BY system_serial_number
HAVING NVL( MAX( CASE l.role_id WHEN 19 THEN l.cmat_customer_id END ), 0 )
<> NVL( MAX( CASE l.role_id WHEN 1 THEN l.cmat_customer_id END ), 0 )
) a
LEFT OUTER JOIN
eim_pr_system s
ON ( s.system_serial_number=a.system_serial_number )
Since your original query is not throwing a TOO_MANY_ROWS exception on the correlated sub-queries, I am assuming that your data is such that there is only a single row being returned for each correlated query and the above query will reflect your output (although without some sample data it is difficult to test).

Answer 2

Apart from 'making the query faster' - there is a way to achieve a faster export using SQL Developer.
When you use the data grid, export feature - this will execute the query, again. The only time this won't happen is if you have fetched ALL the rows into the grid. Doing this will for very large data sets will be 'expensive' on the client side, but you can avoid that.
For a faster export, add a /*csv*/ comment in your select, and wrap the statement with a spool c:\my_file.csv - then collapse the script output panel, and run that with F5. As we fetch the data, we'll write it to that file in a CSV format.
/*csv*/
/*xml*/
/*json*/
/*html*/
/*insert*/
I talk about this feature in detail here.