RE: Security Use Requirements

[Michael Eisler <mre@zambeel.com>]

> If the trade-off is the expense of required physical resources to implement
> privacy, then mandating presents of these resources does not allow the

What physical resources, a few kilobytes of code?

> customer a choice of whether they wish the expense.  Should the

Not mandating the presence of those resources results in interoperability
problems, and impacts customer choice more severely. As Mr. Robinson pointed
out, the lack of mandatory to implement security services in NFSv[23] resulted
in only a handful of implementations supporting such services. So the customer
has no choice about whether to use stronger security flavors that a minority
of NFS implementations support, unless he wants to limit his vendor list to a
handful. Hence my rhetorical question: "Why not allow the customer to make the
tradeoff between security and performance?"

The question for the WG should be whether privacy is critical or not. If is
critical, then privacy must be mandatory to implement. If it is not then the
specification had better make convincing arguments why it isn't critical,
otherwise we can expect IESG or the Area Advisor to reject the specification.
Been there, done that.

I find it hard to believe that privacy in ips is not critical. Much of the
data being rendered is rather valuable.

	-mre