Rethinking Security: Seven Steps To
Tighten Network Security
by Alan Mark
Since the events of September 11, 2001, people
everywhere have been evaluating what security means to them. Many
people are afraid to fly because they think it is too risky. After
all, if you are a frequent flier, you know the challenges that
officials face when trying to secure an airport.
Other people have avoided visiting major cities
because of possible terrorism. Some people have been reluctant to
attend sporting events such as the World Series. These are expected
reactions for the general public.
In the IT world, we too have been rethinking how
security affects our lives. On September 24, the New York
Times reported that the terrorist attacks also "destroyed
crucial links to state computers that manage welfare, Medicaid and
food stamp cases, leaving thousands of poor people in New York City
and nearby counties without normal access to emergency cash, food
and health care." (See Nina Bernstein, "Destroyed Computer Links
Leave Thousands of Poor People Without Welfare Benefits," New
York Times, Sept. 24, at http://www.nytimes.com/.)
Clearly, the information highway is just as important as the
buildings that house the records.
Security has always been an integral part of the IT
infrastructure, and like airport officials, we face enormous
challenges in securing our company's networks. Security threats can
be both electronic and physical. Electronic threats include hackers
and intruders who attack web sites, e-mail systems, or in-house
software. Physical threats include rogue network administrators,
employees, and contractors as well as fires, bombs, floods, and
earthquakes.
When a security breach is discovered, losses follow
even if no damage is incurred. For example, if a security breach
occurs in an airport, all passengers are evacuated, possibly causing
delays in the whole air-transit system. If a company discovers an
electronic break-in, that company may shut down its Internet access
to prevent further intrusion. Of course, shutting down Internet
access also stops electronic business transactions from customers,
vendors, and remote offices. How can we as IT professionals minimize
security violations and the resulting losses?
RESPONDING TO THREATS
Our response to traumatic events is not always
logical: For example, pocketknives, nail files, and other small
devices are now banned on commercial flights. In fact, metal cutlery
knives have been banned from all airplanes and airports. Passengers
receive a metal fork and a plastic knife with which to eat their
meal (even at airport restaurants). Restricting a cutlery knife
wouldn't have prevented the tragedies on September 11, but any knife
is now perceived as dangerous.
In the IT world, some software products have continual
security vulnerabilities. For example, the Code Red worm, which
exposed a security vulnerability in Microsoft Internet Information
Servers (IIS), cost companies an estimated U.S. $2.62 billion,
according to researchers at Computer Economics. (See "Economic
Impact of Malicious Code Attacks" at http://www.computereconomics.com/.)
Although Netcraft estimates a slight drop in the usage of Microsoft
IIS servers in the fall of 2001 (after the Code Red worm had hit),
more than three million Microsoft IIS servers were still in use in
November 2001. (Netcraft estimates there were 300,000 less Microsoft
IIS servers in use in November than there were in October. For more
information, see http://www.netcraft.com/.)
The level of risk we take in all areas of life depends
on what the perceived threats are and how we can respond to those
threats and recover from any subsequent damage. Clearly, we must
reevaluate security that seemed acceptable before September 11. We
have entered an era where simple passwords and tape backup systems
don't provide adequate security. Now more than ever, we must design
computer systems that minimize security risks and create plans that
can be activated if a disaster strikes.
When setting up or evaluating security, security
professionals adhere to three important principles:
- Security is only as good as the weakest link in a chain of
systems. You must implement strong security policies for an entire
system. Otherwise, security policies are pointless.
- Security is never 100 percent. In the real world, you cannot
secure everything. However, you must continue to implement and
test new procedures to attain the highest level of security
possible.
- Someone must be trusted. For example, having military
personnel with machine guns guard airport-screening areas is only
effective if the military personnel have gone through rigorous
background checks. In the IT world, network administrators must be
trusted to maintain systems, to control access to data, and to
protect the internal workings of the IT infrastructure.
How do you ensure that network administrators can be
trusted? You can create a circle of trust and human redundancy,
whereby a group of network administrators is responsible for
maintaining systems and these administrators periodically
cross-train one another. You should also have network administrators
periodically rotate tasks. In this way, if one network administrator
does something out of the ordinary, another administrator will
probably discover and report the violation.
No one is immune to sabotage and disasters. Even
before September 11, the White House was the victim of a
denial-of-service attack with the Code Red worm. Ask yourself a
difficult question: If you can't fully secure your company's network
all at once, what steps should you take to begin the process?
This article explores the first steps you should take
to tighten security on your company's network and to minimize
security vulnerabilities.
STEP 1: EVALUATE YOUR COMPANY'S
ENVIRONMENT
The phrase, "if it ain't broken, don't fix it" doesn't
apply to the IT world. If you followed this principle, you would
never apply security patches or install virus software as long as
users could still access their applications. Obviously, you cannot
ignore network security until that security is compromised.
The most difficult aspect of implementing security is
defining it. The basis of any security program involves three
principles: integrity, confidentiality, and availability. In other
words, a security program should 1) prevent data from being
maliciously manipulated, 2) allow data to be seen by a select group,
and 3) make data accessible to only those who need that data.
Several companies specialize in full-service security
audits (also known as vulnerability or risk assessments). These
Managed Security Service Providers (MSSPs) can evaluate nearly every
aspect of your organization's IT environment, including identifying
who can physically access critical data systems. (For fun, you may
want to rent the movie "Sneakers" to see how Robert Redford's
company evaluated a bank's security system.) The cost for this
service can range from tens of thousands of dollars to hundreds of
thousands of dollars, and the evaluation can take several months to
perform.
Is such an audit worth the money and effort? As in
your personal life, you must assess what your company's assets are
worth. Of course, you must factor in items that have no calculable
value, such as photos or other items that cannot be replaced. For
example, if you bought a home for U.S. $150,000, you would purchase
that much insurance for it. If your deductible were U.S. $500,
adding a U.S. $300 alarm system would make sense. More importantly,
the alarm would protect uninsurable items.
How do you evaluate how much your company's data are
really worth? Unfortunately, it's difficult. Imagine if you suddenly
lost your e-mail address book. How long would you spend recreating
the list? Or what would happen if your company's sales database
system were infected by a virus? If the backup took one day to
restore (not unreasonable given the size of today's databases), how
much revenue would your company lose?
You also can't easily compute a return on investment
(ROI) from implementing most security solutions, except those
solutions that add functionality to an existing system. For example,
if a home alarm included a camera to check your premises via the
Internet, that alarm would serve a double benefit. In the business
world, where reducing support calls results in cost savings, you may
have an easier time selling a single sign-on solution to upper
management.
You can't put a dollar figure on vulnerability because
it's merely a threat. Are there still safe towns where people don't
lock their doors at night? In the Internet age, your company's doors
are always open, and the entire world is just outside.
Although you can't easily determine how much your
company's data are worth, you can determine what would happen if
your company's data suddenly disappeared. In your evaluation,
consider a worst-case scenario (such as a fire and complete
destruction) and a more likely scenario (such as sabotage by
intruders and hackers).
Depending on your company's business, another area of
concern may exist: industrial espionage. Although a typical hacker
(also known as a cracker or script kiddie) wants to be
seen and heard, an industrial intruder's goal is to steal
information without detection. The industry does not have hard
evidence on how much spying occurs because many thefts go unnoticed.
Evaluating your company's environment can be an
overwhelming task. The best approach is a step-by-step analysis,
beginning with a current diagram of your company's network
architecture. The "Checklist
for Evaluating Your IT Infrastructure" will help you evaluate
entry points and critical processes on your company's network. This
checklist will also help you address security issues such as
determining how confidential data is accessed and who can access it.
Perhaps the most overlooked part of a security
analysis is writing policies. For example, you should write policies
to ensure that software patches are installed, only authorized users
can enter the data center, and smart cards are required to use the
accounting system.
In fact, ensuring that software patches are installed
is critical. I am amazed to read about companies that connect
unpatched systems to the Internet. Patches often fix security
vulnerabilities. If you do not apply patches, your systems may have
gaping security holes. Scanners find these vulnerable systems and
alert hackers.
You should not connect any system to the Internet
unless that system is fully patched. To ensure that the software on
your company's network is completely up-to-date, you should maintain
a good relationship with your vendors and review service level
agreements (SLAs) with them.
Policies are both electronic and written. You can use
ZENworks for Desktops or a similar product to enforce electronic
policies, such as users can access only certain applications from
certain workstations. You must write policies to inform employees
that they should not share passwords or use another person's
workstation without first authenticating to the network or to the
workstation. (For more information about implementing security
policies, read "Generally Accepted System Security Principles" at http://web.mit.edu/security/www/gassp1.html.)
As we learned from the disasters of September 11,
police and militia alone cannot secure the world. Everyone must be
involved and alert. You should explain to employees how important it
is for them to speak up if something does not seem right or if they
discover a security flaw. You may even want to set up an anonymous
mailbox for users to write suggestions or send information about
security issues.
You should also warn users about the danger of e-mail
messages. Just as the U.S. Post Office advises people not to open
letters from unknown senders, you should advise users not to open
e-mail messages from unknown senders--especially e-mail messages
that have attachments. Because the temptation to open these messages
can be too great for some users, you may want to filter suspicious
e-mail messages at the gateway before they reach users.
Another important system to implement is electronic
auditing. Some excellent products monitor both NetWare and
eDirectory: Visual Click's DSMeter and DSRazor (http://www.visualclick.com/),
Blue Lance's LT Auditor+ (http://www.bluelance.com/), and
NetVision's DirectoryAlert (http://www.netvision.com/). In
addition, AdRem Software's NetTrend (http://www.adremsoft.com/)
monitors many network devices.
These products generate detailed reports about nearly
any activity that occurs in eDirectory or on a server. For example,
using one of these products, you can see which users have
administrator rights and if these users' rights are hidden through
access control lists (ACLs). You can also generate reports on
password settings and failed logins to eDirectory.
In particular, DSMeter logs eDirectory and NetWare
file system security changes, and DSRazor audits eDirectory security
and includes more than 100 predefined reports. DirectoryAlert and LT
Auditor+ feature real-time intruder alerts and secure audit trails.
Implementing one of these systems is crucial to knowing what is
happening on your company's network.
STEP 2: IDENTIFY YOUR USERS
The basic part of any computer system is identifying
the user. The main problem is that each system implements a
different procedure to identify a user. As a result, user BOB in one
system may be different than user BOB in another system.
For years, companies have explored consolidating user
accounts, without much success. Unfortunately, most systems do not
communicate with each other, and companies may not be able to
replace legacy systems with newer standards-based applications. So
what's a poor network administrator to do?
For starters, you may want to try Novell access and
security solutions, which coordinate the tedious task of creating,
deleting, and identifying users. Novell access and security
solutions enable organizations to securely manage access to
applications, databases, and platforms for web, wireless, client,
Virtual Private Network (VPN), and dial-up users. With these
platform-independent solutions, users have only one set of
credentials to access all of the network resources to which they
have rights. And, of course, it doesn't matter which platform these
resources run on. (For a list of these solutions, see "Novell
Access and Security Solutions.")
The identification process in any system ranges from a
simple username and password to complex biometric IDs. Remember the
secure lab in the first "Mission Impossible" movie? In that movie,
the U.S. CIA used a variety of sensors, biometric, and smart card
techniques to protect a super-sensitive system.
In the real world, of course, most companies can't
afford such fancy techniques. However, relying on only a password to
protect critical information isn't adequate anymore. Fortunately,
you can improve the methods you use to identify users.
Identifying a user can include three basic login
factors: password (something you know), token (something you have),
and biometrics (something you are). Combining these factors results
in a multifactor authentication, which essentially proves a
user's identity. A clearance level (also called a
grade) is assigned to the user, and data are restricted based
on this clearance level.
For example, you could set up a system so that a user
who authenticates with a username and password can access only basic
applications and data. If a user authenticates with a fingerprint,
however, that user can access more sensitive information such as the
payroll system.
Novell Modular Authentication Service (NMAS)
Enterprise Edition uses Novell and third-party modules to implement
any or all of these login methods. (NMAS is a Novell access and
security solution. For more information about NMAS modules, visit www.novell.com/products/nmas/partners.)
The login methods you choose depend on how much trust you require to
access a resource, the cost of implementing the method, and whether
or not the method is used for other functions. You can choose from
the following options:
- Password. The NDS password has been the basis of
authentication since 1993. Two other password options are
available: the simple password and the enhanced password.
The simple password stores another encrypted phrase in
eDirectory and is used by various Novell products (such as Native
File Access Pack). The simple password can also be used for
Internet-related functions. The NDS password, on the other hand,
is never sent across the wire and is inaccessible to applications.
(For more information, visit http://developer.novell.com/research/appnotes/1994/october/02/04.htm.)
The enhanced password allows you to set up rules for the password
(such as passwords must include upper- or lowercase letters,
numbers, or special characters).
In addition, a Personal
Identification Number (PIN) can be used to unlock smart cards or
can be combined with token devices to identify a user.
Passwords are well-suited for web-based authentication
because no special hardware is required. However, passwords are
more easily stolen and hacked than other methods, and they can be
written down.
- Token. Most tokens are double-factor based: They
require something that you know (password or PIN) and something
that you have (the device). Some smart card readers also have a
fingerprint reader for added protection (biometric). (See Figure
1.)
Several types of tokens are available today. The
most popular is SecureID from RSA Security Inc. (http://www.rsasecurity.com/).
This time synchronous token displays a new six-digit code every 30
seconds. The user enters the code plus a PIN to authenticate.
In a similar fashion, Vasco (http://www.vasco.com/) offers
Digipass, a token that displays a six-digit number but only after
the user enters the correct PIN. The device offers a
challenge-response option that doesn't require a precise clock in
the server. Like SecureID, Digipass is well-suited for Internet or
intranet use.
Rainbow Technologies Inc. (http://www.rainbow.com/) offers
a small USB token that stores users' credentials and certificates.
A password unlocks the token for use. Because the USB token
requires special software to be installed on the PC, this token
won't work at Internet cafes.
Most smart cards operate in
a similar fashion. The user puts the smart card into the reader,
unlocks the smart card with a PIN, and the credentials are used to
authenticate the user.
Smart cards and USB tokens have two
advantages over other devices: The workstation can be locked when
the device is removed, and a private key stored on the device can
be used to digitally sign a transaction (such as an e-mail
message).
A new device, called a proximity card, is
also available. A sensor is attached to a workstation, and this
sensor automatically recognizes the card the user is carrying as
he or she approaches. To prevent a thief from automatically
authenticating with a stolen card, an additional form of ID is
required (usually fingerprint). This solution is being implemented
in some hospitals for fast terminal access.
- Biometric. How does the non-computing world identify a
person? By the way he or she looks. In small organizations, people
easily recognize their colleagues. In larger organizations or in
public, however, people rely on ID cards.
ID cards have
three problems: First, people's appearances (such as hair color,
beards, wrinkles) change over time so ID cards must be updated.
Second, ID cards may be easy to forge. Third, photo ID cards don't
translate well in the electronic world.
Obviously, we need
a different approach for network IDs. Biometric techniques have
been around for a while, but until recently, they weren't
reliable. Today, biometric devices can accurately identify a user
by his or her fingerprint. (See Figure
1.) Other methods, such as voice and face recognition, are
also possible. (See Figure
2.) SAFLINK Corp. (http://www.saflink.com/)
supports a variety of biometric devices (www.saflink.com/bsp.html)
and uses NMAS as the authentication platform.
Of course,
the trick is that the biometric reader cannot produce false
positives. In other words, the reader cannot mistakenly identify
you as someone else and authenticate you with the wrong ID.
TLC Care Hospital in Las Vegas uses a hand-geometry device
to authenticate employees when they start and stop work. This
system ensures that no one can punch in or out for another person.
In addition, 55,000 employees at Chicago airports will soon use
fingerprint identification to access secure areas. This
fingerprint identification will be provided by SecuGen, another
NMAS partner.
STEP 3: EFFICIENTLY MANAGE USER ACCOUNTS
Most users have multiple user accounts. For example,
suppose that Alice has an Alice Windows account on her workstation
and logs in to eDirectory as Alice. However, Alice's e-mail ID is
AliceW, and when Alice accesses a UNIX-based payroll application,
she's known as AliceWinters. Alice's mainframe ID is AW09379 (her
initials and employee ID). Of course, each account has a different
password. What's Alice to do?
You know only too well that multiple accounts cause
headaches for you and the users' whose accounts you manage. Lost
passwords mean more telephone calls to the help desk, resulting in
U.S. $50-$80 per incident according to IDC. (For more information,
visit http://www.idc.com/.) And
what happens if "Alice doesn't live here anymore?" How quickly can
you disable all of her accounts?
Old systems may fade away, but they never die because
they're too costly to change. So you're stuck trying to teach new
tricks to old systems. Fortunately, several solutions can help you
manage user identities on multiple platforms.
The first solution is NDS Authentication Services (http://developer.novell.com/nss_profile.jsp?product_key=79958).
This solution synchronizes passwords between the following systems,
even if the usernames are different:
- AIX
- FreeBSD
- HP-UX
- Linux
- OS/390
- Solaris
- Windows NT/2000
With NDS Authentication Services, Alice uses the same
password---her eDirectory password---to access her applications,
regardless of the platform on which the applications are running.
Some applications can also access eDirectory user properties, such
as group membership and security equivalence.
Another solution, Novell DirXML (www.novell.com/products/nds/dirxml),
synchronizes not only usernames and passwords but also other
user-related data. For example, when a user's address and phone
number are changed in an Oracle database, Novell DirXML can send
that change to eDirectory and other databases and programs such as
your company's e-mail system. As a result, the change is made only
one time, and all of your company's databases and addresses books
contain current information.
In fact, eDirectory and DirXML are the foundation of
Novell's Zero Day Start initiative. This eProvisioning solution sets
the standard for synchronizing disparate systems in an organization.
With Zero Day Start, new Novell employees can be productive the
first day they start work. Rather than entering employees'
information in multiple systems (a process that could take days or
even weeks), Novell can enter this information only once. The
information is then synchronized with the appropriate systems.
Equally important, this solution makes Zero Day Stop a
reality. You know that quickly removing or disabling user accounts
can be a nightmare. With Novell DirXML, you can disable user
accounts with the click of a button.
In addition, NetVision's Synchronicity (http://www.netvision.com/)
manages multiple Windows NT domains by using eDirectory as the
synchronizing mechanism and single point of reference. NetVision
also recently announced NetVision Policy Management Suite, which
automates the process of synchronizing passwords on disparate
systems. This suite includes a library of scripts that help enforce
security policies on those systems.
STEP 4: PROTECT PASSWORDS
Face it: most systems--UNIX systems, mainframes, and
web sites--use passwords as their only method of authentication.
Despite all that's been said about tokens, smart cards, and
biometrics, the unforgiving weak password is probably still your
worst nightmare.
Where do users keep their list of passwords? This list
may be written on a sticky note and placed under a keyboard or on a
monitor. It may even be stored on a laptop or hand-held device. If a
simple solution to the password chaos isn't available, users will
creatively devise methods of their own--even at the risk of having
their passwords stolen.
The answer to this password dilemma is Novell
SecureLogin, which protects your organization from the plethora of
problems related to the unsophisticated collection of characters
known as the password. First, Novell SecureLogin provides single
sign-on services to web, host, Citrix sessions, and Windows
applications. Second, Novell SecureLogin incorporates
directory-based password policies, which you create. Finally,
because the user's credentials are stored in the directory, they are
accessible from any workstation that is running the Novell
SecureLogin software.
With Novell SecureLogin, dozens of applications are
predefined for automatic password detection. In this case, Novell
SecureLogin automatically stores and remembers the user's
credentials with minimal user interaction.
If the application is not predefined, Novell
SecureLogin includes a wizard to help the user set up a single
sign-on application. For web-based logins, Novell SecureLogin asks
the user whether or not it should remember the user's credentials
(username, password, and other data).
The traditional concern over single sign-on services
has been that one password unlocks all other passwords. If the
eDirectory password is stolen or hacked, someone may be able to
access a mission-critical application because the username and
password are automatically entered. To prevent this problem, you can
use NMAS to require stronger eDirectory authentication. Before
Novell SecureLogin enters the user's credentials for the
mission-critical application, another form of ID must be verified
(such as a token or fingerprint).
What if the user authenticates to eDirectory and
leaves to get a cup of coffee? The workstation is now open to any
passer-by. At least two solutions are available: You can configure
NMAS to lock the workstation after so many minutes of inactivity, or
you can use a smart card or token to secure the workstation. If this
smart card or token is removed, the workstation is automatically
locked.
This last alternative requires the employee to take
the token when leaving the workstation. Attaching the token to a key
chain or photo ID card helps ensure employee compliance.
I speak to thousands of network administrators every
year on various security topics, and I am amazed that many don't use
a single sign-on product. Such a product not only increases password
security but also saves you time. Try it out; download a demo
version of Novell SecureLogin from www.novell.com/products/securelogin,
and use it on your own workstation in standalone mode. On my own
workstation, I have defined more than 50 credentials to applications
and web sites.
STEP 5: SECURE WORKSTATIONS AND SERVER
CONSOLES
Today's workstations store more data than servers did
five years ago, and therein lies the problem: When storage is
readily available on drive C, users have no reason to save data
elsewhere.
Storing data locally presents two problems: Data
probably won't be backed up, and data is easily accessible if the
workstation is stolen or hacked.
Novell's ZENworks for Desktops (www.novell.com/products/zenworks)
uses eDirectory to enforce workstation policies. (See Figure
3.) With ZENworks for Desktops, users and workstations become
manageable entities and are associated with inheritable policies.
For example, you can create a policy that forces a virus scan and
backup of a workstation--whether or not the user is logged in to the
network. With Wake-Up On LAN technology, a PC can be powered on via
remote commands.
In addition, ZENworks for Desktops policies can
dynamically create eDirectory users on Windows workstations. A
user's profile is then copied from the user's home directory to the
workstation, and another policy dictates what the user can see and
do.
For example, after Alice authenticates to eDirectory,
her desktop environment and preferences are created on the
workstation. A ZENworks for Desktops policy prohibits Alice from
accessing the Explorer, Run, and Registry commands, thereby
preventing her from accessing external applications. When Alice logs
out, her profile is updated on the server and removed from the
workstation, preventing anyone else from accessing Alice's account.
The key to ZENworks for Desktops is eDirectory, which
is the keeper of user and workstation information and the policies
that affect them. If Alice's eDirectory account is disabled, she
won't be able to log in to any workstation.
ZENworks for Desktops also includes the Application
Launcher, which installs, distributes, launches, configures,
repairs, and uninstalls nearly any Windows application. The most
important security feature of Application Launcher is that if a user
doesn't have rights to the application, that user can't even see the
icon to double-click it. Hackers have a difficult time breaking into
systems they can't see.
Of course, you should also secure the server console.
The traditional RCONSOLE utility is considered unsafe to use because
passwords aren't securely encrypted. To solve this problem, AdRem's
sfConsole 4.0 provides secure access to both local and remote server
consoles. With sfConsole, you simply authenticate to eDirectory and
establish a secure 128-bit key connection.
In addition, sfConsole 4.0 safeguards the server
console with a password-protected screensaver and keyboard blockade.
You can also grant certain users limited rights to the server
console, and you can specify which commands these users can execute
at the server console.
STEP 6: SECURE DATA
Securing data is a complicated undertaking. After all,
data has to be usable, which means it can be printed or viewed on a
monitor. With networks, information is also easily copied, and if
everyone has access to the network, an eavesdropper can capture data
sent across the wire.
You can take two approaches to securing data: You can
prevent unauthorized people from accessing data, and you can encrypt
data in case an unauthorized person gains access to it.
In the NetWare environment, you can place access
control on volumes, directories, and files. In eDirectory, you can
restrict which objects or attributes can be seen or modified.
However, if an intruder steals a password--or maybe even a
server--data are exposed. You must encrypt data over network wires
and on the hard drive itself.
As mentioned earlier, NMAS can add a security label to
a NetWare volume, requiring all users--even the ADMIN user--to
authenticate a certain way before they can view data stored on the
volume. For example, if the FINANCE volume has a Password&Token
label, a user would need at least that clearance level to see stock
information. This clearance level means the user would need to use a
token such as a smart card and PIN. With NMAS, how users
authenticate determines what they can do. (For more information
about using clearance levels, see "Taipei
County Government Secures Access to Its Assets.")
Novell iFolder (www.novell.com/products/ifolder)
safely synchronizes data between workstations and servers. Files
stored in a specified folder are encrypted before being sent to the
server. Even if someone were to steal the server, the encrypted data
are worthless to the thief. However, users with the necessary rights
can securely access files from any web browser. iFolder uses Secure
Sockets Layer (SSL) to protect the transmission. (For more
information, see "Novell
iFolder: Your Data Where You Want It, When You Want It,"
Novell Connection, May 2001, pp. 6-20.)
Encrypting data is an effective way to thwart thieves.
But how do you know that the data hasn't been tampered with?
Encryption keeps the data safe; digital signatures help ensure data
integrity.
When an e-mail message or file is digitally signed,
the recipient can be sure that it came from the sender and wasn't
tampered with. The biggest problem with encrypting or signing a
document is getting the credentials (public keys) of the sender and
recipient. The details of digital signatures are beyond the scope of
this article, but in general, in order for Alice to send an
encrypted message to Bob, she needs Bob's public key. In order for
Bob to verify that the message came from Alice, he needs Alice's
public key.
GroupWise 6 includes new features that help solve the
problem of finding a user's public keys. Before Alice sends her
message, GroupWise prompts her to use a private or public LDAP
directory to find Bob's key. Once found, the certificate (including
Bob's public key) is stored in Alice's address book. (See Figure
4.)
If you are looking for software to secure e-mail
messages, PGP Security Business, a division of Network Associates,
encrypts and signs files and e-mail on multiple platforms. A huge
database of users freely uses the software. (For more information
about PGP products, visit http://www.pgp.com/.)
A new approach to data encryption is to require two or
more authentication factors to encrypt files, folders, or an entire
hard drive. After the user authenticates to the local system, the
user's passphrase is sent to the encryption software. PC Guardian
(http://www.pcguardian.com/)
has several products on the market to protect data, and this company
is developing a version that works with eDirectory.
Although laptops make working on the road easier than
ever, they are the least secure piece of equipment an organization
owns. Hundreds of thousands of laptops are stolen each year (1 in 14
of all manufactured), and I'll bet that most of the data are
unprotected. (For more information, visit www.ztrace.com/zLab1.asp.
For statistics on lost computers, visit www.safeware.com/losscharts.htm.)
Even if the laptop operating system requires a
password, a hacker with the right tools can eventually break it. If
the laptop isn't backed up, data are lost with the hardware. Using
ZENworks for Desktops to set up workstation policies that require
laptops to be backed up and using Novell iFolder to synchronize data
between workstations and servers can make laptops much more secure.
As you know only too well, viruses, worms and other
nasty creatures can infect workstations and possibly make them
unusable. Worse yet, viruses and worms can send sensitive data to a
host on the Internet. Viruses infect systems in three basic ways:
through a security hole in client software (such as Microsoft
Windows, Outlook, or Word), by gaining access to web services
(through Microsoft IIS or a UNIX-based application), or via e-mail.
Again, you can use ZENworks for Desktops to force
regular virus scans, and your security policy can explain why users
should not open e-mail messages or attachments from unknown senders.
You may also want to scan e-mail messages at the gateway.
STEPS 7: SECURE INTERNET ACCESS
Managing internal users and systems can be difficult,
but when users need to access internal systems via the Internet, you
probably get justifiably nervous. And when vendors and customers
also need access, you may feel like opening your medicine cabinet
before proceeding.
Your first line of defense is the so-called
firewall. Yes, that one system will do everything from
preventing disasters to keeping the bad guys out. It's a floor shine
and a shoe polish!
I'm sorry to be so sarcastic, but firewall is
the most misused word in the networking world. No one device can
protect an entire network, nor should you rely on one device to do
this. You should think of a firewall as a collection of systems,
services, and policies that protect your internal network from
intruders and data loss.
A firewall solution, therefore, encompasses
software and hardware found on servers, Internet gateways, and even
workstations. A complete firewall solution should include
intruder-detection systems, authentication controls, auditing
monitors, access management to hardware and data centers, virus
software, and policies to enforce employees' compliance.
Before opening your network's floodgates to the
outside world, you must determine which systems must be accessible.
You will probably discover that nearly all of your company's systems
must be accessible to Internet users. Because many systems are now
or will be web-based, you need to identify users via a web browser
interface.
Novell's iChain puts a front-end on corporate
resources by requiring users to authenticate by password, token, or
certificate before those users access back-end systems. In addition,
HTTP data is automatically encrypted within an SSL session
established by iChain. As a result, all communications between the
Internet user and the back-end web server are encrypted without
changing the web server.
BorderManager provides another type of secure
transaction with a VPN. Using VPNs, each client establishes a secure
connection to a server. After authentication, users access the
intranet as if they were in the office. Server-to-server VPNs
encrypt data between servers using the Internet as the transport
medium.
Data must be secured at all points. Although securing
Internet transactions is important, protecting the back-end systems
is imperative. There has not been one proven case of a credit card
number being stolen in a single transaction. However, there have
been many reports of intruders entering databases that contain
credit card numbers. Both iChain and BorderManager can effectively
block access to back-end systems.
PUTTING IT TOGETHER, BIT BY BIT
With eDirectory as the foundation of your company's
network, you know who a user is (identity), where that user is
(location), how that user logged in (authentication), and what that
user can see (access control): The following real-life scenario
shows the power of using eDirectory with other solutions mentioned
in this article.
Bob arrives at his desk and logs in to eDirectory. His
account is valid, so his profile is transferred to his workstation.
It's Monday morning, and a new virus was discovered over the
weekend. The ZENworks Application Launcher configures Bob's Start
menu and taskbar, downloads the latest virus signature file, and
begins running a virus check. Because Bob is in the Sales division,
a ZENworks policy allows nearly full access to his workstation
(although he cannot use Regedit).
Bob launches a terminal emulator to access a sales
database. His mainframe ID is BobT. He enters his username and
password (which was synchronized to his eDirectory password).
Bob next wants to check the financial system to see
sales figures for his area. The financial system requires a
biometric clearance, so Bob reauthenticates using a fingerprint
reader on his keyboard. When the Application Launcher refreshes, the
financial system icon appears on Bob's desktop.
Now Bob wants to check his stock holdings. He launches
a browser and goes to the web site. Novell SecureLogin automatically
enters his username and password for the web site. A free golfing
lesson offer is shown in a pop-up window, and Bob clicks on it.
However, he can't access that web site. The network administrator
used BorderManager to block sports-related sites.
Bob then launches GroupWise. Novell SecureLogin
automatically enters his GroupWise username, but a dialog box
appears, telling Bob that it's time to change his password. Novell
SecureLogin automatically helps Bob create a new password based on
rules defined by the network administrator.
Bob composes a message to his boss. Before sending the
message, he encrypts and signs it. Now the message will be sent
confidentially, and his boss can prove it came from Bob.
At home, Bob accesses the corporate intranet by first
authenticating to the Novell iChain server. Thereafter, all
communications between his home computer and the intranet are
secured via SSL, and sign-on to corporate web applications are
handled automatically.
CONCLUSION
Let's face it: security isn't sexy. Most security
projects tighten the control over users, workstations, and
buildings. For that reason alone, cracking down on security worries
some network administrators because they have to appear as bad guys.
Your organization will always be subject to security
threats, both internal and external. Your job is to minimize the
risk. Remember that security is an on-going process. Continually
update your systems with the latest security patches, continually
assess your environment for risks, and implement the best security
systems your company can afford.
Finally, don't forget your most important resource:
people. Users can be difficult, but they can also be your first line
of defense. Ask them to be part of the security process. For
example, while I was flying back from Comdex in November, a flight
attendant asked me if I would be willing to help the crew if a bad
situation occurred. In this new era of unexpected situations,
everyone must be involved.
Alan Mark is the chief security strategist for
Novell. He has worked at the company for 11 years and speaks to
large organizations around the globe.
Novell Connection, December 2001, pp.
6-23 |