I have a question on flex table performance for loading data.
we are dealing with a flex table with around 200 columns for loading CEF data for which we are using fcefparser(). We are using two cluster node to utilize most of the CPU, we are loading data in parallel using multiple copy command process. And we are dealing with 2 files per process per node to process.
The command looks like,
copy CEF from local '/home/sharedlog/Sample_50k.cef', '/home/sharedlog/Sample_50k - Copy (2).cef' on v_fs1_node0001 parser fcefparser() DIRECT;
likewise we are issuing 8 commands in parallel through vsql script.
But if we install database on a single node and then if we test the same scenario then the result turns out to 10 times faster than two node cluster.
Generally clustering should help increasing performance in addition to high availability. So, why adding node slows down a loading speed on a node?