SAP HANA | BULK INSERT in SAP HANA - bulkinsert

Question

We have SAP HANA 1.0 SP11. We have one requirement where we need to calculate current stock at store, material level on daily basis. No of rows expected are around 250 million.
Currently we use procedure for same. Flow of procedure is as follows -
begin
t_rst = select * from <LOGIC of deriving current stock on tables MARD,MARC,MBEW>;
select count(*) into v_cnt from :t_rst;
v_loop = v_cnt/2500000;
FOR X in 0 .. v_loop DO
INSERT INTO CRRENT_STOCK_TABLE
SELECT * FROM :t_rst LIMIT 2500000 OFFSET :count;
COMMIT;
count := count + 2500000;
END FOR;
end;
Row count of result set t_rst is around 250 million.
Total execution time of procedure time is around 2.5 hours. Few times procedure goes into long running state resulting into error. We run this procedure in non peak hours of business so load on system is almost nothing.
Is there a way, we can load data in target table in parallel threads and reduce loading time. Also, is there way to bulk insert efficiently in HANA.
Query for t_rst fetches first 1000 rows in 5 minutes.

Answer 1

As Lars, mentioned the total resource usage will not change effectively
But if you have limited time (non-peak hours) and if the system configuration will overcome to the requirements of parallel execution, maybe you can try using
BEGIN PARALLEL EXECUTION
<stmt>
END;
Please refer to reference documentation
After you calculate v_loop value, you know how many times you have to run following INSERT command
INSERT INTO CRRENT_STOCK_TABLE
SELECT * FROM :t_rst LIMIT 2500000 OFFSET :count;
I'm not sure how to convert above code into a dynamic calculation for PARALLEL EXECUTION
But you can assume let's say 10 parallel processes, and run that many INSERT command by modifying the OFFSET clause according to calculated values
The ones that you exceed will run for zero rows which will not harm the overall process
As a response to #LarsBr. , as he mentioned there are limitations that will prevent parallel execution
Restrictions and Limitations
The following restrictions apply:
Modification of tables with a foreign key or triggers are not allowed
Updating the same table in different statements is not allowed
Only concurrent reads on one table are allowed. Implicit SELECT and SELCT INTO scalar variable statements are supported.
Calling procedures containing dynamic SQL (for example, EXEC, EXECUTE IMMEDIATE) is not supported in parallel blocks
Mixing read-only procedure calls and read-write procedure calls in a parallel block is not allowed.
These limitations saying, insert to same table will not be possible from different executions and dynamic SQL cannot be used too