Optimise Load via Proc Casutil

[avinashjha1787]

I am trying to load a MS-SQL server table to CAS via Proc Casutil Load. The table has more than 100000 records with 2 columns of type NVARCHAR(MAX) which are really big string variables. The load table takes around 4hrs to complete. My environment has 5 worker nodes each of 1TB.

What are the options I have to speed up the load ?

[Patrick]

You could use numReadNodes=0 to ensure you're using all your workers for multi-node data transfer and then also look into sliceColumn= and sliceExpressions= in order to get evenly distributed data.  

[Patrick]

And I forgot to mention readBuff= . You might have to do some testing to figure out the optimal number of rows to fetch in one go.

[Mazi]

Do you need all 100K+ rows?

Though this doesn’t seem much IMHO.

You might want to look at PROC CAS
Table.loadTable action as you can add a where clause to this and other parameters that are not available to PROC CASUTIL.