Tài liệu Windows 2000 Security docx

It also reviews the new Windows 2000 security protocols, andprotection of services and data.. The Need for Security If you are new to network administration in general and Windows 2000 a

Windows 2000 Security

This chapter starts you off with a discussion on the need

for powerful distributed security before introducing you

to the specifics of Windows 2000 distributed security services

It also reviews the new Windows 2000 security protocols, andprotection of services and data

Windows 2000 Security

While the new era of computing and Windows 2000 will bringforth many benefits, it will also herald dastardly attempts torob you, beat you up, and shut you down There are manyforces out there that have only one thing on their evil minds,and that is to find any way to break into your network to plun-der and pillage

Before you start building your new corporate infrastructurearound Windows 2000, it will pay for you to become thor-oughly versed in the security mechanisms the operating sys-tem offers and how to go about locking down your assets

Without a doubt, it is probably the most secure operating tem available today Not only has it inherited the Windows NTC2 security compliance, which was a ton of work for Microsoftand set the stage for a secure Windows 2000, but also, if therewere showbiz awards for security, Windows 2000 would clean

sys-up at the Oscars, the Golden Globes, the Grammies, and more

But before we get into Windows 2000 security specifics, let’slook at the problem holistically, then you can evaluate yourcurrent security status before devising a security plan

You have probably heard the term everywhere, so what doesC2 security mean to you, the network or server administrator?

Absolutely nothing C2 security is nothing more than a U.S

government sanction The United States keeps a series of

“books” that grade the security levels of operating systems

Windows NT passed with distinction because it was able to

Logon andAuthentication

demonstrate compliance of the C2 specifications These specifications includeobject ownership, object protection, audit trail, memory protection, and user identi-fication, all of which are discussed in various places in this book.

C2 is defined in the so-called “Orange Book,” which is really titled the Trusted

System Evaluation Criteria C2 evaluation checks to see how secure a computer

really is However, C2 only applies to standalone computers Microsoft is also testing to the specifications for network computers (Red Book and Blue Book).Microsoft has gone above and beyond C2 with Windows 2000 So the term is really meaningless

The operating system is not C2 out of the box Everyone has access to everything

A vendor or security service provider has to set up a machine and the OS to be compliant This means locking down objects, setting up audit trails, creating useraccounts with secure password philosophy, and so on Only when a machine hasbeen fully locked down can it be rated as C2-compliant no matter if it’s a wash-ing machine or a file server

C2-C2 security meant a lot to Windows NT, and whatever hoops and hurdles Microsoftwent through and over to gain C2 security is not lost in Windows 2000 However,

we are now playing away from home the field is the Internet, and the game is e-commerce You have high-powered security protocols to configure, and you have lots more room to drop the ball

Another reason that C2 is not important to you is that, as mentioned earlier, out ofthe box Windows 2000 is as locked down as the space above your head You have

to lock down every aspect of it; the network is only as secure as you make it IfWindows 2000 is not properly configured, claiming awards like C2 will not get youout of a jam when a hacker pulls your pants down on the Internet We know we arebeing blunt, but security is part of the day-to-day life of a network administrator Ifyou don’t have a security problem, you don’t have a network

The Need for Security

If you are new to network administration in general and Windows 2000 (and NT) inparticular, then before you devise a security plan, you need to understand the risks

to your network and yourself Unless you plan to hire a security expert, you willprobably have to come up with a plan yourself Chances are your company will askthis of you your superior will assume that you are well versed in the subject Ifyou are well versed in the security threat, you can skip this part and go directly tothe section titled “Rising to the Challenge.”

A company’s data is its lifeblood, and it needs to be vigorously protected As thenetwork administrator, you will be required to ensure that data is kept confidentialand that it can be relied upon There are numerous mechanisms in place to assistyou with respect to data integrity and confidentiality, and they range from sensibleaccess control policy to encryption, backup, and availability

Note

Data Input

Data is vulnerable to attack and capture from the moment a person types in a user

ID and password How often have you had to enter a password while someone wasstanding over your shoulder? You try to type as quickly as you can, but spies willwatch you typing and pick up your passwords quicker than you think Then, whenyou are not at your desk, they will get your user ID from the memo field at the sign-

in screen and masquerade as you from any computer, anywhere

The new smart card technology has been introduced in Windows 2000 and is cussed later in this chapter With a smart card, the user is authenticated withoutrisking being compromised because the thief needs the card to complete the hack

dis-Smart card readers offer one of the most sophisticated domain authentication tions available to Windows 2000

solu-Data Transport

The PC’s or input device’s operating system must transport the information downthe network stack to the transport, all the way to the domain controller’s (DC’s)network interface and up the DC’s respective stack All along this route, the data isvulnerable to interception If the data is not encrypted, or is encrypted very lightly,there is a risk that a person tapping the network will be able to pick up conversa-tions between your input device and the domain controller, or any other partner for that matter

To counter this, Windows 2000 employs extensive encryption technology both indata and network communications, and in file storage and protection

Why the Threat Exists

There are many reasons people threaten your security Let’s look at a short list ofthreats that you are most likely to encounter during your life as a Windows 2000Server administrator:

1 Espionage: People need to break into your communications realm to learn

com-pany secrets, employee secrets, product plans, financial situation, strategy, and

so forth This level of threat is the most virulent The attackers have strongmotives to get the attack under way and to ensure they succeed The attackers

do not want to be discovered and will continue to hide in your environment aslong as they need to The damage is often irreparable if the attackers are undis-covered This is the most difficult form of attack to counter because, for themost part, you do not know where they are hitting you or why

While bugging devices and spying are not usually the responsibility of the work or server administrator, espionage via the network is becoming moreprobable every day because it is so easy and it is where all the jewels arelocated

net-Over the network, hackers will read files and e-mail, and try to log in to bases wherever they can to steal credit card numbers, bank account numbers,and so forth An attacker can, for example, find out the password of yourvoice mail system and then listen to your messages.

data-2 Denial of Service (DoS): These attackers are intent on destroying you They

can attack your physical premises or locations, which is becoming harder to

do all the time, or they can target your network, which is becoming easier to

do because you are connected to the Internet or because you provide userswith remote access This is fast becoming the favorable means of attack forstopping your work: firstly, because of the dependency your company has onthe network, and secondly, because the attacker does not need to be physi-cally present for the attack

DoS attacks are made by flooding your network portal (targeting your gateway

to the Internet) with massive floods of e-mail, or with syn attacks, which are

the low-level communication barrages that suck up all the server’s resources,finally causing it to crash Sometimes the objective is to crash the server just

to trigger backdoor code that spawns a process There could be a millionplaces on a network to hide a sliver of code that gets executed when certainfiles are loaded Good examples are the boot files and startup files likeAUTOEXEC.BAT

3 Hostile Applications: Hostile applications are placed on the Internet for

unwary surfers to download Upon execution of the code on your internal network, the application can begin its dirty work, which for a while might be

to do nothing that can cause it to be detected, but rather to find informationthat would be valuable to the attacker Such applications are also calledTrojan horses

4 Virus Attacks: By far, the most visible attack on the network comes in the

form of viruses Contrary to the claims that there are tens of thousands ofviruses, only a handful of virus writers can actually claim to have inventedone from start to finish Most virus authors are not as brilliant as you mayhave been led to believe; they are just copycats However, this informationdoes not provide any relief

A lot of virus code is available on the Internet to be freely downloaded, ulated, and enhanced or packed with a payload This is the reason we see somany variations of viruses every month Some can be detected by anti-virussoftware such as NetShield and cleaned up; others are more sinister, such asBackdoor-G, which can only be picked up by the anti-virus software after ithas delivered its payload Not only does it wreck your PC before it can bedetected, but it also first attacks the anti-virus software

manip-Threats emanate from two locales: the external environment and the internal ronment These two environments can be easily defined as follows:

envi-✦ The external environment: The threat comes from people who have no

con-tractual status with the enterprise They are complete strangers The attackcomes from the outside

✦ The internal environment: The threat comes from people who have a

rela-tionship with the company, from employees to contractors to customers

The attack usually comes from the inside In some cases, it comes from theoutside, with inside information Other times, the threat is not born out ofrevenge or criminal intent, but ignorance

The External Environment

Not too long ago, the only way to threaten or attack an organization, its people,

or its business was through some sort of physical act This is no longer the case

It costs far less money and is much safer for a hacker to stay in a safe haven andattempt to break into a network through a RAS portal or connection to the Internet

For many, it means the possibility of financial reward; for others, it has to do withsome form of demented feeling of achievement

Now that many small companies can afford dedicated connections to the Internet,the pickings have become very attractive While we have not yet realized the paper-less office, almost all data is placed on the network in share-points and databases

The network and server storage silos are thus loaded with valuable information

Attackers also no longer need to proactively choose their targets They create tile code that gets inadvertently downloaded from the Internet and gets executed

hos-by a number of mechanisms, from rebooting to the mere act of unzipping a file

The code then can gather intelligence and send it to its master It is therefore tial that you establish policy to ensure that code downloaded from the Internet isauthenticated and signed with the digital signature (a public key) of a trusted soft-ware publisher

essen-E-mail is now very much tangible property, and it can be used in court cases as dence and as a source of information that can be used to plan an attack on a person

evi-or an evi-organization We all communicate mevi-ore by e-mail than we do by snail mail, yete-mail is treated like a postcard We do not enclose our messages in an envelopeand seal it We just put it in the mail for anyone to look at

E-mail needs to be secured on two levels We need to be sure that the people withwhom we communicate are really who they say they are And we need to be surethat our e-mail is not being read or changed as it traverses the net It is very easy

to trace the route a message takes over the Internet and penetrate e-mail systems

Securing e-mail is becoming essential and falls under the auspices of public keyencryption, discussed shortly

The Internal Environment

The internal environment threat comprises employees who are either malicious,stupid, or who make honest mistakes Threats come in the form of outright misuse

of privileges to total ignorance or stupidity For example: The perpetrator of right misuse of privileges has administrative rights on the network and provideshim or herself access to sensitive data

out-The ignorance factor often involves users failing to keep anti-virus software current,

or downloading all forms of rubbish from the Internet, thereby introducing cious content to the network from the external environment

mali-Outright stupidity and honest mistakes that often cause headaches for tors are usually deleted files, corrupted databases, deleted mailbox folders, and thelike Deleted data can usually be recovered from backups, as long as the backupregimen is well practiced in your company Most of the time, recovering deletedfiles is just a waste of time spent doing administrative work to have to keep recov-ering files Often, the problems are not user-related issues at all, but just bad man-agement on the part of a lazy network or server administrator

administra-Rising to the Challenge

Over the years, there has been a lot of discussion about the security capabilities ofWindows NT Microsoft has often been criticized for not delivering a more secureoperating system when, in fact, the opposite is the case But it has not been allMicrosoft’s fault For starters, the U.S government has for years not allowed theexport of 128K-bit encryption algorithms although that did not deter many orga-nizations from smuggling out the software

And as for the comparison with UNIX, UNIX systems are more at risk today thanWindows 2000 Since the UNIX source code is open for all to see, many hackers canread the code to look for weak points and plot their attacks Server for server, thereare still more UNIX machines on the Internet than Window NT or Windows 2000machines On Windows NT, hackers resort to scanning network communications tolook for information with which to replay attacks Data interception was and still is

a common form of attack against an NT network

For Windows 2000 to compete and even excel over the competition in the risky and

exposed world of e-commerce, it needed to be the most secure operating system.

The following sections explore the standard Windows 2000 security mechanismsMicrosoft has implemented in Windows 2000:

Note

Before you tackle the protocols, you need to get up to speed on the dagger stuff.

cloak-and-Encryption 101

This is a true story A man walked into a diner one morning and ordered fried eggs

When the eggs were delivered, he changed his mind and advised the waitress that

he had ordered scrambled eggs The waitress, peeved at the cheek of the client,picked up a fork and with a quick whipping movement rendered the eggs into anunrecognizable heap “There, now they are scrambled,” she said, and stormed off

The action of rendering the eggs into an unintelligible mess is known as scrambling

Data is scrambled in similar fashion; we call it encryption At first, the data is in

whole recognizable form, often called plain text, like the fried eggs The motion to scramble them is known as the algorithm and the result is often termed cipher

text In the anecdote, the algorithm is the technique, style, or “recipe” by which

the waitress used her wrist and fork to turn a perfect pair of sunny-side-ups into amound of yolk and white If she only took a few stabs at the eggs, the patron might

be able to claim he still had fried eggs (not a strong encryption algorithm)

Knowing the key that reverses the process is vital to the recovery of the data, butthat is the only difference between egg scrambling and data scrambling If we knewhow to unscramble eggs, Humpty Dumpty might still be alive, and our world would

be very different

In computer science, the standard that governs the techniques and recipes forencryption of data is known as the Data Encryption Standard (DES) DES dataencryption algorithms (DEAs) specify how to encrypt data and how to decrypt that data A number of important bodies, such as ANSI and the National Institute

of Standards and Technology (NIST), govern the specifications for DES Each rithm is rated according to the strength of its encryption ability (and resistance

algo-to duplication, attack of the encryption/decryption key)

DES, actually the DEAs, needs to be continuously improved because the codes areoften cracked by encryption experts (for science and crime) New standards are onthe horizon, and soon the Advanced Encryption Standard (EAS) will replace DES

Other standards governed by these bodies include the Digital Signature Standard(DSS) and the Digital Signature Algorithm (DSA) Incidentally, the U.S governmentdoes not regulate encryption

For more information on encryption standards, see the RSA Laboratories Web site

at www.rsasecurity.com

Note

Cryptography

Cryptography dates back more than 4,000 years Over the past millennia, it has protected many a culture’s communications and has brought them through wars,treaties with neighbors, and more

In recent years, electronic data communications have escalated to such volume andimportance in our lives that without electronic or digital cryptography we wouldnot be able to continue on our logical course

In fact, we owe our computerized environment to cryptography If you have timeduring the locking down of your networks, you should read the biography of AlanTuring, who directed the British to build the first digital computers to break theGerman’s Enigma code

Pretty Good Privacy (PGP) is a software program written originally and distributedillegally for no financial gain by Phil Zimmerman, who believed that the cryptographyalgorithms that were being protected by patents should be made public property worldwide He created PGP back in 1991, and over the years, it was disseminatedaround the world on the “undernet.” Even though its export was expressly forbidden

by the U.S government’s International Traffic in Arms Regulations, which classifiedhis software as a munition, it became available everywhere on bulletin board systemsand the first pioneer sites of the World Wide Web In the last decade, PGP was prettymuch the only means of securing data and communications on the Internet and cor-porate networks of the world

But encrypting data always required a user to make an effort to secure tions Lethargy and lack of knowledge have always left room for error and holes.Only with the incorporation of the encryption algorithms in the very core of theoperating systems and standards-based network protocols would encryptionbecome as pervasive and as transparent as air

communica-We have come a long way since Phil Zimmerman risked detention to make the

slo-gan encryption for everyone a reality Today, Windows 2000 incorporates it

exten-sively Only you, the administrator, need to ensure that it is configured correctly,through security policy, and everyone on the network will be able to use it, withouteven knowing it exists Before we look at this native support for cryptography inWindows 2000 and how it is used, here is some cryptography 101

Keys

Cryptography is a lock, a means of securing information by rendering it pherable without a key The key, or cryptographic key, is held closely by peoplesending and receiving the communication The following is the simplest example

undeci-of cryptography:

The communication: Package color baby burgerThe Key:

Package = meet color = same baby = grand central station burger = 14:00 hours

Deciphered: meet me at the same place at Grand Central station at 2 p.m

Obviously, if you have the key, you can unlock the code and decipher the message

Public Keys

Public key encryption uses two keys One key is public, and the other is private.

Both keys can encrypt data, but only the private key can decrypt the data To bepervasive, the technology depends on a public key infrastructure (PKI), whichWindows 2000 now supports (more about PKI later)

A mathematical process is used to generate the two keys, and the keys are related toeach other by the product of that mathematical process So the message encryptedwith one key can be decrypted only with the other This is how it works:

You want to send an encrypted message The receiver has a public key, which he

or she makes publicly available for encrypting messages You encrypt the messageusing the public key and send it When the receiver gets your message, he or shecan decrypt it using the private key, which is mathematically related to the publickey No one, including you, can decrypt the message with the public key

It goes without saying that the private key must be closely held or your messageswill be compromised

Session Keys

The chief problem in making public keys widely available is that the encryptionalgorithms used to generate public keys are too slow for the majority of just-in-timecommunications (there are numerous algorithms used to create the keys, but thetechnology is beyond the scope of this book) For this reason, a simpler session key

is generated, and it in turn holds the “key” to the encrypted data

1 A session key is randomly generated for every communication that requires

encryption A key distribution authority (or the originator of the

communica-tion, or a vouchsafe process) creates the session key for the communication

or message

2 The data is encrypted with the session key.

3 The session key is then encrypted with the recipient’s public key The

encryption of the data by the session key is a thousand times faster than the encryption of the data by the public key

4 The encrypted data and the encrypted session key are then sent to the

receiver, who can decrypt both by first decrypting the session key with the secret key and then decrypting the data with the session key

Key Certificates

Key certificates are containers for public keys Key certificates usually contain the

public key of the recipient, the identity of the creator of the public key, the date the key was created, and a list of digital signatures

Digital Signatures

We sign most things we do in the material world, so why not in the digital world?Most of us spend our working lives in cyberspace Our customers deal with us onthe net, they buy from us on the net, and they expect that when they send us confi-dential communications, they are sending it to the right people We also want toknow that when someone sends us a message, hits our Web site, or connects to ourcomputers that they are who they say they are We also need to use digital signa-tures to prevent repudiation In other words, if someone places an order with youover the World Wide Web or via e-mail, or enters into some form of contract withyou, they should sign the document so that they cannot turn around later and repu-diate the transaction

It is also not always necessary to encrypt a message, which taxes computerresources Sometimes, the message or data content or information is not sensitive.Sending someone a publicly available encrypted price list would be an absurd idea.But what if someone intercepted that message and changed the content, whichwould affect the relationship? What if someone sent you a message saying, “Maryjust had a little lamb,” and a jokester intercepted the message and changed the con-tent to read, “Mary just ate her little lamb?” The effects could be devastating

Digital signatures are thus used to authenticate the sender, to legally bind parties

in digital transactions, to authenticate content, and to be sure that content has notbeen changed or tampered with in any way

Windows 2000 makes wide use of the encryption mechanics described above One

of the most important implementations is in the use of the Kerberos protocol,which is now the most important means of authentication and protection of data

in not only Windows 2000, but also all major operating systems

Kerberos

What if we told you that every time you come to work you have to go to a certainsecurity officer who signs you in and issues you a clip-on tag that allows you toenter the building and go to your desk, but do nothing else? And that you had tocheck in with the officer every hour to renew your tag?

What if you then needed to go to this person for a new tag every time you needed

to access a resource in the company, such as the file room or the copier machine?

And then what would you think if we told you that you have to present this tag toguards that protect each resource so that they can verify that you are legitimate?

You’d say, “Wow, this is overkill Why is security so tight here?” It would probably

be hard to work in such an environment But what if several companies, or a wholecity, adopted such stringent security practices? Life in the city would be so securethat companies would be able to trust each other enough to share resources Butfor all intents and purposes, it would still be hard to work in such an environment

Yet, this is precisely how Kerberos works The only difference is that the securitycheck-ins and tag issues are handled transparently by the underlying protocols, and everything takes place in network transmissions The user is oblivious to what

is going on under the network hood

Kerberos is based on a system of tickets, which are packets of encrypted data that are issued by a Key Distribution Center (KDC) — the security officer we just men-

tioned This ticket is your “passport” and carries with it a myriad of security

infor-mation Each KDC is responsible for a realm, and in Windows 2000 every domain is

also a Kerberos realm Also, every Active Directory domain controller (DC) is a KDC

When you log on to Windows, WinLogon and LSA kick in to first authenticate you

to the KDC (see Chapter 2), which provides you an initial ticket called the Ticket

Granting Ticket (TGT), which is akin to a right-of-way coupon at the fairground, or

a passport Then, when you need to access resources on the network, you presentthe TGT to the DC and request a ticket for a resource This resource ticket is known

as a Service Ticket (ST) When you need access to a resource, your processing

envi-ronment presents the ST to the resource You are then granted access in dance with the ACL protecting the resource

accor-The implementation of Kerberos in Windows 2000 is fully compliant with theInternet Engineering Task Force’s (IETF) Kerberos v5, which was originally devel-oped by MIT This specification is supported by many, which means that ticketsissued in a Windows 2000 domain (now also known as a Kerberos realm) can bepassed to other realms, such as networks running Mac OS, Novell NetWare, UNIX,AIX, IRIX, and so forth.

Trusts can therefore be established between the Kerberos Domain Controllers(KDCs) in the respective realms The KDC trusts, for all intents and purposes, workjust like trusts for Windows NT systems, which are set up between the primarydomain controller (PDC) in each domain And because Windows 2000 still speaks

NT LAN Manager (NTLM), trusts are maintained to legacy Windows domains

Kerberos, however, does require more tweaking and administration than you may

be used to on Windows NT domains using NTLM That’s because users have tocheck in with the KDC several times a day For example, if you are logged on for

12 hours straight, you will probably have to check in with the KDC about 12 to 15times in that period If the domain supports 1,200 users, that will result in about18,000 hits to the KDC

Also, trusts between heterogeneous networks are not as transparent as the trustsbetween Active Directory domains, in which the domain controllers can explicitlyvouch for the users Trusts between Windows 2000 forests, Windows 2000 andWindows NT, and Windows 2000 and other realms involve manual setup betweeneach domain’s or realm’s respective administrator The process that takes place inthe UNIX or IRIX realm may be very different to the setup that takes place betweenWindows 2000 realms

When planning the physical layout of the network, if you have multiple domains thatcommunicate across a WAN, you will need to establish shortcuts or the best possi-ble routes that ticket transmission can use to move from realm to realm Shortcutsmay be required so that authentication does not become bogged down in networktraffic over a small pipe

If authentication is slow due to slow links between networks, you may have agood reason to establish the site as a new domain For more information ondeciding when to create a new domain, check out Chapter 7

Kerberos is, however, a very fast protocol and is an ideal environment for menting the Single Sign-On paradigm in network authentication

imple-Kerberos and the Single Sign-On Initiative

Single Sign-On is long overdue From a security angle, it provides tremendous fits If a user has six or seven passwords, it means he or she has six or seven moreopportunities to compromise security Many people are so sick of the different pass-words they have to deal with that they would rather not have a password This is aproblem in systems where the password creation and application is in the hands

bene-Note

of the user A good example is a voice-mail system Many, such as CallXpress forWindows NT, ask the user not to enter 1234 or to leave the password blank But areview of the password history on the system usually shows that many passwordsare left blank or are simply 1234

Other users go to the opposite extreme and type their passwords into a passworddatabase or a spreadsheet, or worse, a simple text file An intruder will go to town

on a document loaded with keys Password databases are the mother lode; it takes

a few seconds to crack the password that locks the file

With Single Sign-On, the user authenticates once, and that authentication isrespected by other network applications and services Made possible by Kerberosand Active Directory, Single Sign-On is supported in SQL Server 2000 and Exchange

2000, and is supported by trusts set up between realms implemented by other ating systems and Windows 2000 It is the very reason that Windows 2000 trusts —between domains that share a common root or forest — are transitive

oper-Psst This Is How Kerberos Works

Kerberos is built around the idea of “shared secrets.” In other words, if only twopeople know a secret, then either person can verify the identity of the other by confirming that the other person knows the secret The shared secret in Kerberos

is between Kerberos and the security principal (the human user or a device).

Here’s an analogy: Two people send each other e-mail regularly and need to be surethat each e-mail cannot be repudiated by the other, or that someone else is not mas-querading as the sender So in order to be sure that the sender or receiver is whothey say they are, both agree offline that something in the messages between themwill confirm that each one is “the one.” However, if someone is analyzing e-mail andspotting word arrangements, it will not take them long to discover the hidden confir-mation message On a network authentication mechanism, this can be quite a prob-lem because it would not take long to intercept a message and fool an authenticationservice into thinking the user is genuine

So how do the two correspondents devise a plan to be certain of their identities?

The answer is symmetric-key cryptography The shared key must be kept secret,however, or anyone will be able to decode the message As discussed earlier, a symmetric key is a single key that is capable of both encryption and decryption

In other words, as long as the two corespondents share the same key, they canencrypt their messages and be sure that the partner is able to decrypt it

The terms secret key and symmetric key are often interchanged when discussing

the use of a single key to encrypt and decrypt text However, it is entirely possiblefor a secret key to fall into the wrong hands

The practice of secret key cryptography is not new and goes back to before theCold War days when insurgents perfected secret key techniques and cipher science

In the Kerberos implementation, however, authentication is a done deal as long as

Note

the information is decrypted, or as long as one party can prove they are the realthing by being in possession of the decrypting key in the first place But what ifsomeone on the network steals the key, or manages to copy previous authentica-tion sessions? Kerberos then makes use of an unalterable factor that goes back tothe Big Bang time.

Time Authentication

Kerberos authentication begins, literally, from the time a user tries to log on to the domain When Kerberos receives an authentication request, it follows thisseries of steps:

1 It looks the user up and loads the key it shares with the user to decrypt the

authentication message

2 It then looks at the information in the message The first item it checks is the

time field, which is the time on the clock of the user’s workstation or machinefrom where the user requested logon authentication If the time on thesender’s clock is out of synch by five minutes, Kerberos will reject the mes-sage without further ado (Kerberos will compensate for the different timezones and daylight savings time) However, if the time is within the allowableoffset of five minutes, Kerberos accepts the message pending one more item

3 Kerberos checks to see if the time is identical or older than previous

authenti-cators received from the sender If the time stamp is not later than and not thesame as previous authenticators, Kerberos allows the user to authenticate tothe domain

However, it is also important to know that the authentication is mutual Kerberos

will send back a message demonstrating that it was able to decrypt the user’s sage Kerberos sends back only select information, the most important being thetime stamp that it obtained from the original authentication from the client If thattime stamp matches the client’s information, then the client is sure that Kerberos,and not an imposter, decrypted the message

mes-Key Distribution

Authenticating to Kerberos works well for authentication to the domain, but whatabout accessing resources once the client has logged in? In that Kerberos is usedfor authenticating to domain resources, how does the client authenticate to othernetwork resources?

Well, Kerberos is able to distribute keys In other words, it acts as a broker This, infact, is where the name Kerberos comes from In Greek mythology, you may recall thatKerberos was a three-headed dog that stood guard over the gates of Hades Kerberos,

the protocol, also has three heads: the client, the server, and a mediator or proxy Theproxy is known as the Key Distribution Center it dishes out keys In Windows 2000,the Key Distribution Center is installed on the Active Directory Domain Controller.

Okay, so now you are beginning to think one step ahead here, and you say, “Cool,that whole rigmarole of decrypting the message and checking the time stamps justhas to be repeated between clients and servers.” And you would be further correct

if you assumed that the job of giving the network resources copies of every user’skey would be that of the Key Distribution Center However, you are correct in the-ory only, because so much key distribution would be a tremendous drain onresources Every server would have to store keys from potentially thousands ofusers in memory What, in fact, is implemented is quite ingenious in its simplicity

Session Tickets

Instead of following the logical plan and sending the session key to the client andthe server at the same time, the KDC in fact sends both copies of the key to theclient and then gets out of the way The client holds the server’s copy of the keyuntil it is ready to contact the server, usually within a few milliseconds The illustra-tion in Figure 3-1 may help you “decrypt” what is going on here

The KDC invents a session key whenever the client contacts it to access a resource(A) The server sends the session key to the client, and embedded in the sessionkey is the session ticket (B) Embedded in the session ticket, which really belongs

to the server, is the server’s session key for the client All that really happens here

is that the KDC acts as a domain broker or proxy for secret key negotiations thattake place between a client and the resource to which it requires access

When the client receives the communication from the KDC, it extracts the ticket andits copy of the session key It stores both items in secure volatile memory Then,when the client contacts the server (C), it sends the server a message that containsthe ticket that is still encrypted with the server’s secret key and a time authenticatorthat is encrypted with the session key The ticket and the authenticator make up theclient’s credentials in the same fashion as the logon authentication process

If everything checks out, the server grants access to the client (D) because theserver knows that a trusted authority, the KDC, issued the credentials As soon asthe client is done using the server, the server can get rid of the session key that theclient was using to communicate with the server The client will instead hold thesession key and re-present it to the server each time it needs to access it

Session tickets can also be reused, and as a safeguard against ticket theft, the ticketscome with expiration times The time to live for a ticket is specified in the domainsecurity policy, which is discussed later in this chapter Typically, ticket life usuallylasts about eight hours, the average logon time When the user logs off, the ticketcache is flushed and all session tickets and keys are discarded