Hi Trafodioneers,

My network skills are very limited, and I was wondering if somewhere in the
communication stack, some encoding can take advantage of transferring data
that is full of 0, vs random bytes.

Like RLE for example…

The reason I am asking, is that part of predicate pushdown V3, the one that
will support align format, I am considering optimizing returned data, and
silence columns that are not required.

For variable length columns that are not needed, that is easy, I will just
return ‘’.

But for fixed length columns, I was wondering if forcing to 0 columns that
we don’t need would result in optimized payload… it would if the network
stack have a layer that is doing RLE type encoding. Is that the case?



Thanks in advance for the help,
Eric

Reply via email to