Re: iSCSI Some Thoughts on Digests

["Mark A. Carlson" <mark.carlson@sun.com>]

Let's be clear as to what we are protecting against.

End to end protection should look at things other than
CRCs to protect the kinds of failures that are prevalent
in other parts of the system. Disk data errors do not look
the same as dropping bits on a serial network connection
(address and data lines "stuck" for example).

The purview of iSCSI, however, is the reliable transport
over a serial network connection and the protection scheme
should be able to detect errors once the data enters the
network. I see no value, therefore, in using CRCs higher
in the stack to protect against these different kinds of 
errors in host memory, DMA transfers or transient disk 
failures that are being done over parallel busses.

-- mark

"Bernard D. Aboba" wrote:
> 
> > Note that the draft needs to insist upon a CPU implementation
> > of the checksum/CRC if it is to be effective.  Implementations
> > that implement the checksum/CRC on some soft or offload
> > processor should be discouraged as this limits the modes
> > of detectable failures.
> 
> Won't this have a very large negative effect on performance?