The
Internet is a two way street
The Internet has been designed in such a way that
all Internet users have equal access to one another. In other words
it is a network of peers, there is no distinction between an Internet
client and an Internet server. When you are accessing the Internet
as a client (e.g. accessing web sites or checking your email) you
may unknowingly be acting as a server yourself.
Each and every Windows workstation has a built-in
file server, which many people use on their Local Area Network (LAN)
to share files with other users. Without the protection of a firewall,
this server may be accessible by other Internet users. All Windows
workstations are now supplied with a "Personal Web Server".
Many users activate this feature in order to build and test their
own web sites. Often, what they don't realise is that this too is
accessible to other Internet users.
Additionally there are numerous bugs and security
flaws in these servers that may allow remote attackers to take control
of your computer with relative ease.
How
will the attacker know that I'm there?
We are all used to finding web sites, either by name
or through the use of a search engine. Underlying this naming system
is the IP Address system. This is the numbering system that allows
clients and servers to target each other. Each Internet connected
computer has its own unique IP address. Those of us browsing on
the Internet don't need to know our IP address, as we are the party
initiating connections to web servers, but we do have one. This
IP Address will accept connections as well as make them.
A Port Scanner is a tool used by hackers and crackers
to find vulnerable computers. They can have it scan thousands of
IP Addresses at random in a matter of minutes. Usually, they're
not looking for a specific target. Anyone will do.
The Internet currently consists of over 4 billion
IP addresses. Hackers and crackers will concentrate on portions
of that address space that are rich in unsecured machines. At any
given time there are thousands of individual crackers running port
scans of thousands of IP Addresses. If you were being port scanned
while you're reading this page, you probably wouldn't know it.
How
will the attacker break-in?
That depends on your system. This is a brief and
far from complete run-down:
Windows 95/98/ME:
The attacker might attack your built-in file server.
Quite often Windows users will share access to their hard drive
on the LAN either without a password, or with a very weak password.
One great feature of Windows 9x/ME from a crackers perspective is
a complete lack of logging. An attacker can try every word in the
Oxford english dictionary as a password, and you have no logs or
alerts telling you that it's happening.
The Personal Web Server of Wndows9x/ME has several
bugs that may allow the attacker to read files from your hard drive
or execute commands. If you are vulnerable to this, the attacker
can download your encrypted password files and crack them at his
leisure. This probably won't even be necessary, as he can now have
your computer download backdoor software such as Back Orifice.
Windows NT/2000/XP:
As with Windows 9x attacks against the built-in file
server are possible. This should be more difficult, as both a login
and password will be required, but every Windows NT/2000/XP comes
with an Administrators account. This is likely to be the target
account for the attacker. The attacker must be more careful than
he would be against a Windows 9x machine, as every failed login
attempt will be logged.
Again the web server will be targeted. IIS is the
NT/2000/XP web server. IIS has a long list of vulnerabilities that
may allow the cracker access to or control of the system.
Linux/Unix/BSD:
The most common attacks against Unix (including Linux
and the BSDs) involve the exploitation of buffer overflows in certain
services. A buffer overflow occurs in response to an attack on software
that contains a certain type of bug (or flaw). Usually the aim of
provoking this buffer overflow is to have the service execute commands
for the attacker.
In recent years there have been bugs in the BIND
DNS server, several FTP servers and the LPR print server. These
packages are often installed by default on Unix systems. An unpatched
Redhat 5.x/6.x/7.x with a default installation will most likely
be vulnerable to one of these vulnerabilities.
Surely
Bug-fixes and good passwords will protect me!
In an ideal world software bugs would not exist,
but they do and for the time being they're here to stay.
New vulnerabilities in software are discovered on
a daily basis. You would have to check the security alert lists
every day to be guaranteed to be up to date.
You can only patch your software against known vulnerabilities.
Quite often the hacking/cracking community know of vulnerabilities
months before the software vendor learns of them. It's not in the
crackers interest to let the rest of us know what he knows.
Minimise
or Eliminate your exposure
Steps to minimise exposure:
- Don't
run unnecessary services. If you don't need
to run file-sharing, web servers and mail servers then remove
them from your system. Unfortunately most operating systems come
with a host of services pre-installed. You should identify which
ones are of use to you and eliminate the rest.
- Firewall
your network. A Stateful-inspection firewall
can prevent unwanted connections from being accepted by your network,
while allowing you full access to the Internet.
- Segment
your network. Exposed services should not
be hosted on the same network segment as non-servers (i.e. workstations).
Your servers are the weakest point in your network. If they are
compromised, the rest of that network segment is sure to follow.
Unlike servers, non-servers can be completely protected from direct
exposure. By placing them in a separate network segment, they
can also be protected from your potentially compromised servers.
- Patch
your servers.
If you must expose services, patch them religiously. Subscribe
to the vendor's security mailing list and other mailing lists
such as Security
Focus's BugTraq.
|