From: Wozny, Scott (US - New York) (swoznydeloitte.com)
Date: Thu Jul 21 2005 - 13:24:42 CDT

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

You're absolutely right. It's part risk analysis, part cost/benefit
analysis. You either choose to accept the risk of pushing out defs
blind because it costs too much in manpower and lost time OR you vet the
sigs and accept the cost of doing so in manpower and that you'll be
exposed for longer but reduce the chance of a repeat of that fateful
Friday, as rare an occurrence as it is, OR you do something in between
that fits for you.

However, _all_ the blame is not on the vendor (though it was a massive
screw-up on their part). There's nothing in that software suite that
_requires_ all defs be pushed immediately, and it used to be that no-one
did. Most of us have just gotten too comfortable with def updates
because problems with sigs so rarely happens. If "other vendors"
patches didn't have so many unforeseen side effects, more people would
push them without testing as well because we're all over worked and we
make those cost / benefit decisions every day.

The concern I had which I wanted to address was with a perceived
implication that it's best to leave AV off IIS boxes (the question this
thread is addressing) because it regularly contains new, possibly
untested code and IMHO that, by itself, does not present a sufficient
risk offset the numerous other benefits AV provides (no matter how
_sure_ you're IIS server is locked down). The event in question had, at
best, a tenuous cause / effect relationship with mitigating factors
which could have prevented it that organizations _chose_ to ignore. It
doesn't matter that everybody does it. Everybody got busted. So we
dust ourselves off and figure out the best way to deal with it. In some
situations, that's to make a conscious choice that enough controls are
in place that AV adds more hassle than it's worth and in _some_
situations that's to take at the servers that are administered by
professionals and put an additional line of defense on them in case
these administrators turn out to be human and make a mistake that AV
might be able to catch. If the term "defense in depth" is unappealing
and too fuzzily defined for you, think of it as "infosec redundancy".
:)

Scott

-----Original Message-----
From: Susan Bradley, CPA aka Ebitz - SBS Rocks [MVP]
[mailto:sbradcpapacbell.net]
Sent: Wednesday, July 20, 2005 7:57 PM
To: Wozny, Scott (US - New York)
Cc: Harlan Carvey; focus-mssecurityfocus.com; jeffshawgo.com
Subject: Re: Should webservers, eg. IIS 6 have anti--virus installed on
them?

Can you honestly say that you vet every dat file that comes your way in
the same manner that your do security patch testing on all of your
systems? Show of virtual hands on this list... how many honestly have
the resources to put the same testbed energy into a/v sig updates as
they do patch deployment? Test it on lab settings/virtual
system/canaries in the office and then roll it out... for all your sized

operations? There are some firms that indeed do this. There are many,
however, that do not. I personally don't have the resources [nor the
a/v deployment set in such a way] that I can do this. Nor do I feel
that the few issues that I have had with allowing a/v to immediately
deploy versus the issues I might have if I don't automate the process
mean that I'm changing my methods.

But...obviously neither did several railroads in Japan, a few Japanese
newspapers and other folks that were also affected and obviously didn't
vet the a/v sigs either.

As often as they are updating these days, the risk of not pushing them
out as they come in has to be weighed with the potential for issues when

not testing them. I'm sorry but this was a A/V dat sig update that
affected the XP sp2 the hardest of all. Trend admitted they screwed up.

As fast as that nailed and flatlined my entire network... there's no way

that should have left Trend's doorstep and been pushed to boxes. It was

an immediate CPU freeze up that had me booting into safe mode to get my
machines back in working mode.

Even Microsoft has expanded their patch testing process to include
external more real life testers. Sorry, but I do not accept that this
dat file freeze up was in any way an acceptable screw up ...and
obviously and unfortunately neither does Wall Street and analysts
...etc....

All I'm saying is we've [I've?] grown complacent and many of us forget
that potentially every hour on the hour new untested code is on our
boxes. Add that to your risk factors and decide accordingly.

Show me an a/v software and this year few of them haven't had their own
security issues as well.

It's called a bit of risk analysis... what's the benefit....what's the
risk. And no matter what size of firm you are... we all play the game,
we just come to different conclusions. Ergo this thread which asked...
what's the risk of webservers having a/v on them?

I think the answer is.. it depends. There may not be a best practice
and instead each one of us needs to perform our own risk analysis and
decide accordingly [I really don't like 'best practices' as a concept
anyway - what's best for me... won't be best for the guy down the
street]

Nah... Dos 5, Wordstar and Lotus 123. Now those were killer apps... I
still have a Compaq Portable luggable in our museum that boots if you
want to try it. In the meantime, excuse me while I go update my
Firefox..again and ensure my Greasemonkey is on whatever version that
isn't vulnerable.

Wozny, Scott (US - New York) wrote:

>Are you actually condemning AV because administrators blindly trusted
>the AV sig updates they received and pushed them to live systems
without
>testing them at all? Who, precisely, wasn't doing their due diligence?
>
>
>Computing is complicated. If one isn't implementing and following
>procedures to protect oneself from screw-ups in other organizations one
>depends upon, then we all really ought to roll back to DOS 6.22 and
stay
>there.
>
>If I misunderstood your implication, please correct me. Otherwise, I
>intend to keep AV in my bag of tricks.
>
>Scott
>
>-----Original Message-----
>From: focus-ms-return-8320-swozny=deloitte.comsecurityfocus.com
>[mailto:focus-ms-return-8320-swozny=deloitte.comsecurityfocus.com] On
>Behalf Of Susan Bradley, CPA aka Ebitz - SBS Rocks [MVP]
>Sent: Wednesday, July 20, 2005 3:32 AM
>To: Harlan Carvey
>Cc: focus-mssecurityfocus.com; jeffshawgo.com
>Subject: Re: Should webservers, eg. IIS 6 have anti--virus installed on
>them?
>
>
>Not to mention ..if you were anywhere near a live system at 3:45 p.m
>Pacific time on a certain Friday when someone didn't do their due
>diligence and flatlined every single one of my workstations and even
>nailed my server....you might make you look at antivirus in a new
>light....
>
>A/V is just introduction of new... possibly untested code on a machine
>.... possibly every hour on the hour....
>
>http://silverstr.ufies.org/blog/archives/000844.html
>
>Harlan Carvey wrote:
>
>
>
>>So far, this is has been an interesting discussion,
>>but beneath it all, I'm seeing what I think is a
>>disturbing trend.
>>
>>
>>
>>
>>
>>>Antivirus needs to be part of the overall security
>>>plan for all Windows machines - it's just part of
>>>the cost of doing business - the cost of the
>>>software, maintenance, and CPU overhead.
>>>
>>>
>>>
>>>
>>I'm seeing absolutist statements like the one above,
>>and it bothers me.
>>
>>If a web server is just a web server, the content is
>>served to the client, going outbound...not coming into
>>the server. If the purpose of the system is to take
>>known-good pages (from the owner) and make them
>>available to the public (over ports 80 and 443), then
>>what is the point of A/V software?
>>
>>I'm seeing a lot of people say that A/V software is
>>necessary, and that it's part of a 'holistic' or
>>'defense in depth' approach, but this really sounds
>>more like Dilbert's "buzz word bingo" than anything
>>else.
>>
>>
>>
>>
>>
>>>Certainly, servers need to be patched, firewalled,
>>>isolated, and locked down. Additionally, code
>>>should be audited for vulnerability to XSS and SQL
>>>injection.
>>>
>>>
>>>
>>>
>>Yes, without a doubt. This is all part of good
>>administration.
>>
>>
>>
>>
>>
>>>None of these things are perfect. Not that AV is
>>>perfect, but it is another layer of defense - making
>>>it part of that "Defense in Depth" strategy.
>>>
>>>
>>>
>>>
>>But, defense against what?
>>
>>
>>
>>
>>
>>>AV has grown into more than just defense against
>>>viruses. It is often effective against worm code,
>>>and some AV has identified common hacking tools
>>>(e.g. - NetCat) as something that doesn't belong on
>>>most systems. You can argue the viability of this
>>>move, but most companies - if they have a security
>>>team - have less that 0.1% of their machines which
>>>maybe should have it there.
>>>
>>>
>>>
>>>
>>"something that doesn't belong on most systems"? How
>>does it get there? If a web server is properly
>>configured and managed, then perhaps the most likely
>>means of infection is from the administrator
>>himself...and in such cases, A/V software is useless.
>>
>>
>>
>>
>>
>>>AV needs to be part of the cost of running Windows -
>>>for better or for worse.
>>>
>>>
>>>
>>>
>>Again, I'm seeing this as an approach that's being
>>parrotted, rather than thought out. I'm not saying
>>that MS products are perfect...not at all. But what I
>>am saying is that using proper administration
>>principles, those that have been espoused for well
>>beyond the past decade, paying additional money to add
>>yet another software package to a web server simply
>>doesn't make good business sense.
>>
>>Why pay more money for another application to
>>maintain, and another set of logs that you're not
>>reviewing anyway?
>>
>>Several years ago, Dave LeBlanc set up an IIS 4.0
>>server in accordance with simple common sense, and it
>>was not vulnerable to Code Red...a full year before
>>Code Red was launched.
>>
>>When Code Red was launched, A/V software would not
>>have helped. However, if the .hta script mapping had
>>been disabled the day before Code Red came out, then
>>guess what? No problems.
>>
>>Should systems have A/V software in place?
>>Maybe...depending upon the function and purpose of the
>>system. Does it make sense? Does it make good
>>business sense? What's the business
>>reason/justification for installing another software
>>package (for $$) over disabling current functionality
>>(which doesn't cost anything)?
>>
>>Harlan
>>
>>
>>
>>------------------------------------------
>>Harlan Carvey, CISSP
>>"Windows Forensics and Incident Recovery"
>>http://www.windows-ir.com
>>http://windowsir.blogspot.com
>>------------------------------------------
>>
>>----------------------------------------------------------------------
-
>>
>>
>----
>
>
>>----------------------------------------------------------------------
-
>>
>>
>----
>
>
>>
>>
>>
>>
>
>-----------------------------------------------------------------------
-
>---
>-----------------------------------------------------------------------
-
>---
>
>
>This message (including any attachments) contains confidential
information intended for a specific individual and purpose, and is
protected by law. If you are not the intended recipient, you should
delete this message. Any disclosure, copying, or distribution of this
message, or the taking of any action based on it, is strictly
prohibited. [v.E.1]
>
>-----------------------------------------------------------------------
----
>-----------------------------------------------------------------------
----
>
>
>
>

--
Letting your vendors set your risk analysis these days?
http://www.threatcode.com

------------------------------------------------------------------------
---
------------------------------------------------------------------------
---

---------------------------------------------------------------------------
---------------------------------------------------------------------------

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]