Johnson, S wrote:
Keep in mind, group policies cannot always be used as in our
environment.
We are a K-12 education and are mandated by federal law to monitor and
protect student access to the internet.
We are now allowing students to bring their own notebooks in on a trial
basis (to be permanent after this summer when we work out the bugs) to
do research on their own computers.
We have to monitor their access to the internet and deny "bad" sites,
again mandated by federal law. So their authentication mechanism is
AD/LDAP to their user ID set up for them to access network resources on
the network.
Since their computers are not on our domain (nor do we want them to be),
we cannot push group policies down to their computer.
In that case your best bet would be to lock down general port-80 access
to them entirely. Using WPAD 'auto-detect' or with students setting
browsers set manually.
That will go a long way toward blocking risky behavior by malware on
mobile devices.
Second best after that would be to setup some helper where they can
authenticate against some other system and the helper permits their
requests past Squid for a time. This provides almost no protection from
malware once the student is browsing a legit session.
Amos
The solution Bluecoat had was very secure, but again their devices are
about $50,000usd / device. As an education provider, that money is hard
to come by especially when we would need 3 devices for the load. Their
authentication mechanism is SOX (sarbane oxley) tested and compliant.
It also works with any computer outbound to the internet. There's no
proxy configuration to worry about; it's all done at the proxy.
Granted, I used WCCP to configure this on Bluecoat which allowed me a
lot of flexibility to add in multiple proxies with ease (and the users
would never know the difference).
sj
-----Original Message-----
From: Kinkie [mailto:gkinkie@xxxxxxxxx]
Sent: Saturday, January 03, 2009 12:51 PM
To: Guido Serassio
Cc: Johnson, S; squid-users@xxxxxxxxxxxxxxx
Subject: Re: NTLM and transparent/interception confusion
On Sat, Jan 3, 2009 at 11:14 AM, Guido Serassio
<guido.serassio@xxxxxxxxxxxxxxxxx> wrote:
Hi Kinkie,
At 18.45 02/01/2009, Kinkie wrote:
Could you try to get a network trace of a successfully authenticated
http transaction?
I would love to see how they do it...
Websense too is using something similar for filtering:
They maintain an IP Address/Username table on the policy server. The
table
can be populated using different ways:
- A logon agent, a little executable running on every client at logon
time
- Direct query to the user workstation
- A DC agent that query DCs for user sessions
There isn't any kind of web browser authentication, and this solution
cannot
work with non Windows clients or machine non domain member.
Multiuser terminal server environments cannot be supported and the WS
policy
server should be Windows based and domain member for full
functionality.
Yuck...
IIRC Squid's "session" helper can do that too then.
This is NOT authentication and it's absolutely insecure: even windows
nowadays supports remote desktops (3 users can share one IP) and SNAT
("connection sharing"), and it's pretty easy to hijack an user's
credentials (simply log on to his workstation as soon as possible
after he's logged out).
an nmblookup-based external authentication helper could be set up to
do one of these, but after all what's the point? If the user has a
proper Windows infrasctructure, it's much easier to use group policies
to configure the browsers..
Thanks for the clarification Guido!
--
Please be using
Current Stable Squid 2.7.STABLE5 or 3.0.STABLE11
Current Beta Squid 3.1.0.3
