On Sun, Jun 10, 2012 at 2:58 AM, Kulin Shah <m.ku...@gmail.com> wrote: > This is more out of curiosity than anything else but why were bit masks not > included as part of the OF1.0 spec to match flows on and only subnets/prefix > masks were included? Was this purely because this would integrate straight > in with existing tcams on existing vendor hardware networking gear or was > there another reason to this. > > > I see a few load balancing for DPI use cases emerging out there amongst > customers who view realize the potential of Openflow and envision leveraging > the power of the protocol but not supporting bit masks to match flows is a > big road block.
OF1.0 was approached from a standpoint of what was 'strictly necessary' for a certain class of applications. So, it's is possible to bit mask source and destination IP addresses, but nothing else. After some deployment experience, it became obvious that it's useful to bitmask src and dst mac addresses, so that feature was added to 1.1, and then general bit masking was added to 1.2. I don't think there was any reason to not include it in 1.0, except it wasn't as clearly useful 2 years ago. - Rob . _______________________________________________ openflow-discuss mailing list openflow-discuss@lists.stanford.edu https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
