Re: IDS event filtering

["M. Dodge Mumford" <dodge () nfr net>]

Billy Dodson said:
> I am wanting to get an idea of what you guys out there filter from your
> IDS sensors.  Some of the sensors I monitor get TONS of events for MSSQL
> control overflows.  If the customer is patched for slammer and does not
> have any SQL services on the internet, is it safe to filter out those
> events?  Do you still want to see that traffic even though you know your
> are not vulnerable?  Thanks!

From a vendor's perspective, filtering becomes rather messy. Some customers
(As Patrick Harper put it) don't want to see IIS attacks against apache
servers (or similar issues). Some customers simply don't want to be told
about Big Issues that were patched long ago. Others want to track and
profile every attacker. 

The question "what is safe" is dependent on the risks that the site deems
acceptable.

I think it's fair to say that most attacks happen over TCP, and, given the
current state of TCP initial sequence numbers, the source will be correct.
However, the attack you've pointed out occurred via UDP and the source was
spoofed. If it's impossible for any machine to be brought onto your network
that would be susceptible to -- or infected with blaster -- then it may be
advisable to disable the check. But if people bring laptops in, the risk
analysis would have to take that into account.

-- 

Dodge

Attachment: _bin
Description: