Lists Home |
Date Index |
On Thu, Apr 14, 2005 at 08:41:54AM -0400, Rich Salz wrote:
> > The
> > issue will be whether the larger message size in the stateless
> > solution will be acceptable or not. How much state are you're talking
> > about?
> Let's assume RSA with a key size of 2K bits, maybe sometimes 4K. A
> signature is the same as the key size, so you're talking 256 or 512 bytes,
> plus the data being signed, of coruse.
> At least one certificate will have to flow in each direction. A
> certificate is signed and has a couple-K of data, so call it 2-4Kbytes
> per cert.
Ok, thanks. I don't know enough about the use cases you have in mind,
nor the security mechanisms themselves to know how genereal an approach
this might be. But assuming 2-4K as a worst-case for the general case,
is it such a big deal? I expect many B2B messages to be an order of
magnitude (or two or three) larger than that in practice.
I can well imagine contexts in which the increase in message size is not
appropriate; "TCWA", The Canonical Web App (i.e. get an HTML page,
display it), comes to mind. But this cost also comes with advantages
too, in particular, for this discussion, security advantages; that
messages whose semantics are functions of information only in the
message, are immune from certain kinds of man-in-the-middle and
subversion attacks. The ability to recover from partial failure -
reliability - is improved too, for the same reason. Plus, as the
message is more self-descriptive, its ability to be archived, used in
long-running asynchronous transactions, etc.., is also improved.
Mark Baker. Ottawa, Ontario, CANADA. http://www.markbaker.ca