From: Faisal Khan (faisal@netxs.com.pk)
Date: Mon Dec 13 2004 - 11:50:59 EST
Miguel,
Thanks for the detailed post. I was just reaffirming my own gut feeling.
Only bummer is doing a good scan on a Class C - takes up a few hours.
Besides the client has lots of security gear, the funny thing is sometimes
the firewalls sense that someone is doing a port scan, other times - it
does not. But the main reason to do it a few times is
(a) To be sure about it - rather than basing an entire pen-test on a single
report of port scan.
(b) Client will most likely ask for multiple scan reports
(c) The IPS devices in their network sometimes will block the port scanner
and sometimes allow it to function based on the originating IP. This is why
we have to resort to multiple IPs from various (read: different) Class C
addresses - in the hope that the IPS devices do not temporarily blacklist
our IPs from where the scan is taking place.
Faisal
At 09:24 PM 12/13/2004, miguel.dilaj@pharma.novartis.com wrote:
>Hi Faisal,
>
>I found that using nmap alone is usually enough, provided you use the
>proper settings. An exception is when you're dealing with a firewall
>trying to assess how exactly things interconnect, in such cases you can
>try hping2/3 or firewalk.
>A short time ago I posted an answer somewhere about the most useful nmap
>settings to scan a "normal" network. IMHO:
>* use a very comon source port, like 80 (-g 80)
>* fragment, and be sure that nothing on YOUR side is trying to defragment
>(-f)
>* use paranoid timing, to avoid overreaction from an eventual IDS (-T0)
>* use SYN scan (-sS)
>* use decoys if overreacting IDS are a concern, and if allowed by your
>contract! (-D {decoy1},{decoy2},...)
>
>Then go for any advanced techniques, as required (for example ACK or
>Window scan).
>You can combine OS detection to the above, scan UDP ports, etc., this will
>depend exactly on the setup of the network you're checking, and what are
>you looking for.
>If you don't know what to expect, scan the entire port range, sometimes I
>found interesting things in high ports (for example a proxy, or a Java
>application server), that were not supposed to be open to the world.
>Lastly, don't forget some of the most esoteric and advanced techniques,
>that are used once every solsctice, like IPID scan from probably trusted
>machines, etc.
>
>Because some times you need to use advanced techniques, very often you
>need to scan more than once, but I also recommend (if possible) to scan
>from a completely different source IP address (example: scanning a certain
>system in Spain from my country showed 2 open ports of a proxy installed
>by the ISP, but these ports were not shown when scanned from the same
>ISP's network).
>
>IMHO nmap is simply the best port scanner out there. But of course other
>people can have different preferences, so no flame wars on port scanners
>please ;-)
>I like it on Linux more than on Windows, *somehow* I found it more
>reliable ;-)
>
>IIRC, Fyodor is a member of this list, so perhaps he can enlighten us all
>(or send us to RTFM ;-)
>
>Cheers,
>
>Miguel Dilaj (Nekromancer)
>Vice-President of IT Security Research, OISSG
>
>
>
>
>
>
>Faisal Khan <faisal@netxs.com.pk>
>13/12/2004 14:46
>
>
> To: pen-test@securityfocus.com
> cc: (bcc: Miguel Dilaj/PH/Novartis)
> Subject: Port Scanning.
>
>
>
>
>What's a good industry practise whilst doing port-scanning during a
>pen-test.
>
>Do you rely on the results of a single vendor's software or do you use
>multiple softwares?
>
>Also, with each OEM/vendor - do you scan once or twice?
>
>I need to do a scan on a Class C Address if that matters in any way.
>
>Faisal
Faisal Khan, CEO
Net Access Communication
Systems (Private) Limited
________________________________
Network Security - Secure Web Hosting
Managed Internet Services - Secure Email
Dedicated Servers - Reseller Hosting
Visit www.netxs.com.pk for more information.
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:54:10 EDT