BCP impact on DB log: per-partition bcp vs full table bcp

[Former Member]

Anyone came across this issue?

I'm running manual sync for DB.  20 threads doing slow bcp, batch size 1k, DB set to trunc on checkpoint, log size is 12GB. 

With 20 concurrent threads copying data (ASE 16 => ASE 15) DB is on the verge of exploding the log - but makes it through.

With 10 threads copying data on full table level and 10 threads copying data on per-partition level the log explodes in minutes.

Anyone has any idea why per-partition bcp may have different impact on DB log?  Sounds completely weird to me but looks like it is? 

Any idea (apart from fast bcp) how to reduce bcp=in impact on DB log?

This is ASE 15.0.3 (hence can't open a case for it) - part of preparation for ase 16 migration...

Cheers,

Andrew