iSCSI: Towards a more effective PDU format

["Robert D. Russell" <rdr@mars.iol.unh.edu>]

All,

Some people are objecting to having 2 (there is no need for SEVERAL,
just 2) header digests.  But to have only 1 header digest for
variable length headers leads to the following problem:

If the header is variable length, how does the receiver find the
header digest WITHOUT using the header data that is supposed to
be protected by the digest (since that data might be unreliable)?

The only answer I see is that the receiver MUST know where the digest
is located in the PDU (WITHOUT looking at the header data first).

And the only way that can be done is to put the digest at a FIXED
offset from the start of the PDU.

>From this it follows that the digest cannot be at the end of a
variable-length header -- the receiver will NOT know where to find it!

However, if the digest is NOT at the end of the header, then 2
possibilities arise:

1) If there is only one digest for ALL the header data, then some
   of that data must FOLLOW the digest.

2) If a digest must always follow the data it protects, then
   there has to be a second digest that follows and protects the
   data not covered by the first digest.


Bob Russell
InterOperability Lab
University of New Hampshire
rdr@iol.unh.edu
603-862-3774