RE: iFCP CRC

[Joshua Tseng <jtseng@NishanSystems.com>]

Matt,

> 
> Joshua Tseng wrote:
> > 
> > Wayland,
> > 
> > One important detail which was accidentally omitted from the
> > iFCP document is that the FC CRC (along with EoF and SoF) is
> > stripped off and not encapsulated in the iFCP PDU.  As you guessed,
> > the reason this was done is because iFCP changes the S_ID and
> > D_ID in the original FC frame.
> 
> Why are the S_ID/D_IDs changed?
> 
> -Matt
> 

S_ID and D_ID can be changed on the FC frame because their values
are only of local significance in the context of an individual iFCP
gateway.  They are mapped by the iFCP gateway to a "remote" S_ID,
D_ID.  When interpreted with the "remote" iFCP gateway's IP address,
they provide the unique address of the remote FC device.  This
mechanism provides for unlimited scalability, as each FC device is
addressed by both IP address and PORT_ID (the FC identifier, not
TCP port).

You might have noticed my earlier message comparing iFCP to NAT.
They are in fact very similar in their mechanisms and goals,
in that both iFCP and NAT improve scalability in an environment
of limited address space.  FC has that problem, where a single-
byte DOMAIN_ID is used for FSPF routing and switch identification,
leading to a HARD LIMIT of 239 switches when you subtract the
reserved DOMAIN_ID's.  By mapping the local S_ID and D_ID to an
global IP address and PORT_ID combination, iFCP thereby provides
for unlimited scalability in the number of FC devices that can be
uniquely addressed.

Hope this clarifies things.

Josh