[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <Pine.NEB.4.58.0309282359421.29519@panix2.panix.com>
From: jays at panix.com (Jay Sulzberger)
Subject: Re: Pudent default security
On Sun, 28 Sep 2003, Paul Schmehl wrote:
> --On Sunday, September 28, 2003 10:20 PM -0400 "security@...enik.com"
> <security@...enik.com> wrote:
>
> > I would add yet another take on this.
> >
> [sniipped a lot of good thinking]
> >
> > I think that the problem is not the protocol or the application. It is a
> > fundamental lack of understanding of the security model and the network
> > as a whole.
> >
> Yes, that is what I was trying to say, however lamely. The preponderance
> of discussions and papers on security today focus on the network and how to
> control the flow of data/packets. But in the final analysis, the problems
> always come down to the individual machine, be it server or workstation.
> Why aren't security ideas focusing on that problem primarily? Oh, we all
> know you shouldn't run unnecessary services, but that's about as far as the
> wisdom goes.
>
> SANS has made some efforts in this area with their best practices
> documents, but where is the software development to address it? The
> Bastille is about the only thing I can think of off the top of my head that
> even attempts to address this area. The OS vendors are beginning to come
> around to the off-by-default model (slowly), but protecting what *must* be
> on (such as CIFS, SMB, NFS) is still a laborious (or outrageously
> expensive) process when you're trying to do it on an enterprise level.
>
> IMO the vendors should be providing these types of tools as an integral
> part of the OS in addition to shipping in an off-by-default model. It
> should be trivial to "do security" in an OS. (It still blows my mind that
> every WinXP box comes with UPnP on by default. RPC I can *almost*
> understand, but UPnP???) I'm saying we need a paradigm shift in *thinking*
> about how an OS should be configured out of the box *and* a paradigm shift
> in the ease of configuration on an enterprise level.
>
> Paul Schmehl (pauls@...allas.edu)
Many computer programs are today:
1. unconscious
2. promiscuous
3. incontinent
4. unsupervised
Most programs should be:
1. somewhat self-aware
2. almost chaste and quite delicate in their affections
3. tight-sphinctered
4. well supervised by programs with the power to detect and suppress bad behavior
oo--JS.
Powered by blists - more mailing lists