RE: Automated IDS response

["Russ Wolfe" <rwolfe () hxcorp com>]

One more point, a FW-IDS system that re-configures itself affords the opportunity for the disruption of legitimate 
traffic by posing as legitimate addresses doing illegitimate things...a whole new spin on DOS ;)

Russ Wolfe
Halifax Corporation



<<< "Marcus J. Ranum" <mjr () nfr net>  2/15 11:52a >>>
Crumrine, Gary L wrote:
> When IDS systems first hit the streets a couple of years ago, I think many
> were caught up in all the GA-GA bells and whistles marketing hype that
> accompanied their release.  After some time to evaluate the products and
> adjust our thought processes on how they are implemented, I think we have
> come full circle on their usefulness and I know we are a lot wiser in our
> implementation.

Amen to that, brother Crumrine! :)

And it's about time, too. A lot of the early IDS' promised things
that were patently ridiculous - kind of like the early generation
of firewalls did. ("If you have a firewall, you don't need to worry
about the security of the rest of your network...")  Now I think
a lot of reality has set in. People have discovered that IDS is a
useful tool if deployed correctly, and that it is valuable for
learning what's going on inside and out of the network, but nobody
expects that it'll somehow act like William Gibson-esque "ICE"
and automatically "heal" a broken network or backtrack and destroy
the bad guys.

> I for one now tend to back off from allowing a product to automatically
> modify my configurations in response to something that has the potential to
> be malicious.  Sounds a bit like jumping at shadows.

Yup.
There's also an analogy here to firewalls. :) Having a firewall
that automatically modifies its configuration is also a bad thing.


> I lean towards manual corrective measures that are a result of some form of
> human thought process and analysis.   Guess I am afraid of AI being
> introduced at this level.  In theory, it sounds like a great idea, but it
> rarely works out in real life.  At least to the degree that would make me
> comfortable with it.  Maybe someday...

If there was "real" AI it would be OK. But I think machine
intelligences won't happen for a while and, if they do, they
will be too expensive to have sitting watching a network. ;)
(If I had a machine intelligence I'd teach it to be a stock
day-trader not an IDS...)

> Right now, I trust a human more than a brick.

Those are very different technologies. ;) You can trust a brick
much better than a human if what you're doing is building a
garage. Humans are terrible structural components. ;) Bricks
are pretty bad network managers, tho.

mjr.