2 weeks ago
I have been looking for a standard approach for loading transnational data from multiple oracle sachems to a single LASR table.
The oracle schemas contains sales data for different tenants and the single LASR table is supposed to contain data from all the tenants so that summary statistics can be calculated. All the oracle table should be able to insert/ update (more than 1000 rows) data to the target LASR table. Data need to be refreshed hourly.
Is there any way so that the LASR table can be loaded in parallel? Whether LASR table partitioning can help ?
2 weeks ago
1,000 rows isn't many. I would just do a normal non-parallel load. What version of VA are you using and is it non-distributed (one app server) or distributed (multiple app servers).
If your VA is non-distributed then check out PROC IMSTAT for inserting into LASR tables:
2 weeks ago - last edited 2 weeks ago
Thanks for the response. By 1000 rows I meant 1000 update/inserted rows per tenant per hour into the LASR table, which comes upto around 20000 insert/update rows per hour into the 150+ gb LASR table.
The system is a 4 node system. VA 7.4
The main concern is if multiple tenant tries to insert/update the LASR table at the same time, what issues might come up and how to avoid them ?
If at all we partition the LASR table by tenant, what benefit could be achieved?