Response to question from Prashanth, Rao

>From:  [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]
>On Behalf Of Rao, Prashanth
>Sent:  Tuesday, January 19, 1999 10:11 PM
>To:    '[EMAIL PROTECTED]'
>Subject:       What is Adaptive detection and response
>>hi,
>i have read the cyberwalls white paper from Aberdeen
group.I n this white
>paper author has mentioned about Adaptive detection and
response (Intrusion
>detection and content filtering).where exactly we need to
implement this?at
>firewall level/router level?or at all levels? or create a
separate layer and
>have a dedicated sever etc.any thoughts and comments is of
great help to
>really implement the perfect enterprise cyber wall.
>prashanth rao

This was forwarded to me for an answer, so I will attempt to
provide one
knowing that I have not seen the whole thread of discussion.
If it is out
of balance, I beg forgiveness in advance. Since the concept
of cyberwalls
is somewhat new, please forgive my wordiness as I want to be
clear in
answering your question(s).

Adaptive detection and response is the concept of tying the
ability of the
IDS network security component to an attack profile and then
adjusting
security policies dynamically in the policy enforcement rule
base to defeat
the attack and potentially add additional security policies
to thwart off
other attack profiles that may be related. Historically,
this has not been
a consolidated operation. Further, this has traditionally
been done on
separate standalone systems offering perimeter defensive
measures and not
ubiquitous throughout an enterprise network environment.

One major differentiator of a cyberwall vs. a firewall is
that a firewall
traditionally segments and protects network-to-network
traffic. Cyberwalls
are components that are installed ubiquitously on a network
- on desktops,
on servers, between trusted network components, etc. Another
difference is
that on firewalls, intrusion detection is usually not
included and not
totally integral to the policy enforcement facilities. In a
cyberwall, IDS
functionality and rule adaption to meet a cyberthreat are
integral
components. Finally, because security policy rule
enforcement engines are
installed on desktops and servers (with full IDS and other
functions), the
overhead and performance matters that typically cause
firewall bottlenecks
on a network are avoided and distributed to the source or
destination
system for the traffic on the network. This means that
performance
requirements are not measured or managed the same way as
traditional
perimeter firewalls: a cyberwall implementation distributes
the performance
burden to many machines and, specifically, to the systems
that are in
communications sessions between each other. In this manner,
performance
characterization is a function of what a desktop or server
is doing as
opposed to what some other machine may be doing that
generates traffic on a
network through a shared resource (like a perimeter or
network-to-network
firewall environment).

Major functions of an adaptive detection and response
include three states:
inspection, detection and protection. We achieve this via:

        - Frame-level capture and full bit-level inspection
of all
          traffic to/from a system or network. On internal
networks, this
          also means ALL protocols - not just IP. On
internal networks,
          multiprotocol operations are the norm and not the
exception.
          While IP is popular and growing, most networks
still have IPX,
          AppleTalk, NetBEUI, SNA, DECnet and many other
protocols that
          can be used to attack systems which are used on
networks. In
          some environments, such as real-time and
manufacturing networks,
          there are many, many custom protocols used to
support the operations
          and these cannot be filtered with traditional
perimeter IP-based
          firewall facilities.
        - IDS functionality at each point of inspection. As
an example, a
          server such as a certificate authority (CA) server
or LDAP server
          have no effective manner, under any operating
system, to detect
          a network attack profile on the server itself.
This is especially
          important on internal network attacks via
intranets and extranets.
          Detection of such attacks allows the individual
system or network
          "sub group" to detect the attack and formulate a
defensive
          strategy on-the-fly. IDS systems that are not
ubiquitous can cause
          performance bottlenecks or lack of timely response
when placed on
          standalone systems in chokepoint configuration
fashion.
        - Frame, packet, application, "stateful" and, in
some cases, proxy
          security policy engine filtering facilities that
not only implement
          standard restrictive access methods, but also
interoperate with
          IDS facilities to properly adapt the operating
security rule base
          to meet the threat being detected. This protection
functionality
          allows a cyberwall to exercise real security
actions dynamically
          instead of just reporting a problem in the
expectation that the
          network or security management team will modify
the rule base
          to meet the threat at some future time.

The best network security implementation for any network is
a layered
security architecture that allows work to be done in a
predominantly
transparent manner to the security entities or applications
installed. In
the case of "ultimate" network security, routers with
security filtering
(like access control lists or ACLs), cyberwalls and
traditional firewalls
co-exist and complement each other's functionality as a
network is secured
in a hierarchical or tiered manner. As each layer of a
network security
architecture is breached, the next layer should inspect,
detect and protect
the next layer. Ultimately, through the implementation of a
layered
security architecture, proper levels of security are applied
to defeat the
level of threat and the appropriate tool at each layer of
network
architecture allows the system(s) on the network to survive
an attack.

Cyberwalls are usually configured much differently, due to
multiple
protocols and differing application and server requirements,
than a
traditional IP perimeter firewall with well-known IP-based
applications. As
a result, there are different management burdens associated
with managing
cyberwalls than traditional firewalls.

Companies will need cyberwalls, firewalls, certificate
authorities, digital
signature servers, router ACLs, token authentication
systems,
challenge-response authentication systems, PKI key escrow
and management,
cryptographic methods and a whole rash of security tools and
systems to
properly secure corporate assets from compromise. There is
no one tool to
solve the problems and no one way to do so. Cyberwalls,
however, introduce
a new network security tool to help solve internal security
problems,
provide comprehensive security management to a high
granularity state and
keep a breached perimeter network connection from becoming
the fatal flaw
in security in a company that potentially kills the
corporate assets that
are unprotected internally.

I hope this addresses your questions. If you have any others
about
cyberwalls, please let me know.

Regards,

Bill Hancock


*************************************
 Bill Hancock, Ph.D., CISSP
 Chief Technology Officer
 Network-1 Security Solutions, Inc.
 The Intranet Security Company
 878 Greenview Dr.
 Grand Prairie, TX  75050
 Tel:   972-606-8200
 Fax:   972-606-8220
 Web:   http://www.Network-1.com
 Email: [EMAIL PROTECTED]
 Nasdaq:  NSSI
*************************************




-
[To unsubscribe, send mail to [EMAIL PROTECTED] with
"unsubscribe firewalls" in the body of the message.]

Reply via email to