Subscribe to Windows IT Pro
August 31, 2000 04:11 PM

Encrypting Internet Email with Exchange Server

Windows IT Pro
InstantDoc ID #15490
Rating: (2)
Lock down an SMTP connection at the server, not at the client

Email creates a new instance of an age-old dilemma: security versus convenience. We face this dichotomy in many aspects of our daily lives. For example, when we leave for work in the morning, most of us lock our homes behind us. Although we'd prefer not to deal with keys and locks, the simple fact is that our possessions would eventually disappear if we left them continuously unprotected. So we put locks on our doors and sacrifice a little bit of convenience for the sake of security.

The reverse—sacrificing security for the sake of convenience—also occurs, specifically in some Microsoft products. Password caching in Windows and Microsoft Internet Explorer (IE) is a prime example. Although many network administrators don't approve of password caching, users' workstations can store passwords so that users don't need to remember them.

Email represents the ultimate in communication convenience coupled with a complete lack of security. The analogy that compares the security of an email message to that of a postcard is absolutely accurate. When SMTP was developed— back when the Internet was a much more trusting environment—security wasn't a major concern. As a result, users today transmit confidential information across the Internet without realizing just how vulnerable that data is.

Suppose two pharmaceutical companies join forces to develop a revolutionary new drug. Scientists at the two companies need to share data and documents concerning their research, and they want to do this via email because doing so is convenient. But what if the scientific documents, as they travel across the Internet in an email message, fall into the hands of a competitor? If that competitor then beats the collaborating companies to a patent on a new drug, the potential lost revenue could amount to billions of dollars.

Obviously, network administrators at the two companies need to secure the email traffic that travels between the scientists. How do the administrators best achieve such security? They can rely on users to encrypt email messages, or they can configure email servers to encrypt messages.

Client or Server Encryption
Microsoft Outlook email clients (Outlook 2000 and Outlook Express) have supported encryption and authentication mechanisms for some time. When you properly configure Outlook, users can simply click a button to encrypt (and optionally sign) a message before sending it. Although this security solution is functional, it has two minor flaws. The first is that the message recipient must have the appropriately configured software (and education) to decrypt the message. The second (and more significant) flaw is that the encryption process depends on an unreliable link: end users, who must remember to encrypt their email. Just as we all occasionally forget to lock our houses and cars, users occasionally forget to encrypt confidential messages.

If you're running Microsoft Exchange Server 5.5 or later, server-based encryption is a better alternative. Exchange Server can encrypt SMTP conversations between mail servers across the Internet. This feature is surprisingly easy to set up and ensures the encryption of all messages sent to a designated domain.

To configure server-based encryption on the sending end, you define specific Internet email domains or subdomains in Exchange Server, then instruct the server to encrypt messages headed to accounts in those domains. For example, I could direct my server to encrypt all messages to the domain ntsol.com before transmitting them.

The administrator of the Exchange server on the receiving end defines a special user account for the encrypted messages. If I've configured my server to encrypt all email going to the domain ntsol.com, I need to ask the administrator of the ntsol.com domain to create an account for my Exchange server to use. Let's say the administrator at the remote site creates an account named DOUG.

At this point, I have simple one-sided encryption, in which all the email I send to ntsol.com is encrypted and all the email that ntsol.com sends me is in plaintext. However, I want encryption to work both ways, so I need to set up a user account in my domain. Then, the administrator at ntsol.com can program his or her mail server to encrypt email that it sends to my domain.

Configuring Encryption
To set up a domain and account for encryption, you must modify settings on the Exchange servers that handle your Internet email. Start Microsoft Exchange Administrator, open the Internet Mail Service Properties page, and click the Security tab, which Figure 1, page 88, shows. By default, the system secures no domains with encryption. Click Add to open the Edit E-Mail Domain security information dialog box, which Figure 2, page 88,shows.

Type the name of the domain you want to secure (in my case, ntsol.com), then select the Windows NT challenge/ response authentication and encryption option. Next, you must define the account credentials your Exchange server will use when communicating with the other Exchange server. In effect, your server must log on to the other server. To define the account your server will use, click Change. In the dialog box that opens, type the NT domain name, account name, and password to use. (I typed INTELLINET for the NT domain and DOUG for the account.) Remember that your system will use this domain name and these user account credentials when connecting to the remote system—therefore, for encryption to work, the remote system administrator must set up the user account you define in this dialog box. After you enter the appropriate authentication credentials for your Exchange server to use, you'll see the NT domain and account names in the Edit E-Mail Domain security information dialog box's Windows NT Account field.

Now, whenever my Exchange server sees an outbound message headed for any email address ending with @ntsol .com, it will automatically encrypt the message. The address might be postmaster@ntsol.com or webmaster@ntsol .com—Exchange Server will encrypt all @ntsol.com addresses, regardless of content. An administrator doesn't need to add new addresses for encryption or install software on workstations.

NT Challenge/Response authentication enables NT LAN Manager (NTLM) encryption for an SMTP session. The strength of the encryption depends on your OS installation. If you have the 128-bit (US-only) version of NT at both ends of the connection, you'll get 128-bit encryption. If you have the 40-bit (exportable) version of NT at both ends of the connection, you'll get 40-bit encryption. If you have a different version at each end, NT will probably negotiate down to the 40-bit level because the configuration dialog box doesn't offer an option to "require" strong encryption.

Server encryption is a great feature if you need to share confidential information with business partners who run Exchange Server. You can secure all the email traveling between your systems in just a few minutes by enabling this option.

Related Content:

ARTICLE TOOLS

Comments
  • Scott
    4 years ago
    Jan 11, 2008

    Let me say this for one, business users are not nearly as vulnerable as one may believe, but still vulnerable none-the-less. This is because most businesses have dedicated connections to the internet, and the connection tunnels through this way. However, home users are especially at risk, since many of them use CABLE INTERNET. This is a shared line, so someone that had the proper equipment, or the know how could easily listen in on the line and intercept anything they were looking for.

  • x
    5 years ago
    Apr 12, 2007

    Carl Campos
    Going to try to keep this simple. There are a couple of ways to do this. One would be how you described it, by compromising physical security. A more realistic approach that I have done during a pen test is to do ARP poisoning. This is also known as a type of MITM (man in the middle) attack. You send out ARP packets to 'poison' the nearest router and the host (or router and router). By poisoning these devices (making them think that they are sending traffic to legit devices/systems when they are not) you are now able to intercept the traffic. Just Google "ARP Poisoning" is you want more clarification.

  • Douglas Toombs
    12 years ago
    Oct 04, 2000

    For those of you who have questions in regards to whether or not data ever gets intercepted as it crosses the Internet, here is the response from Mark Joseph Edwards at windowsitsecurity.com:

    "When you send data across the Internet, those packets travel over many networks. Anywhere along the route data can be captured. If someone wanted your data through PacBell, they'd probably crack a system and install some sniffer software--that way they don't have to physically break in. This can and does happen all the time. There are countless news reports on the Internet of data being stolen in this exact manner."

    I hope this information helps. Thanks for reading!

  • John Taylor
    12 years ago
    Sep 26, 2000

    I would just like to add my support to Carlos Campos´comments if anyone can offer an explanation I would be very grateful.

  • Carl Campos
    12 years ago
    Sep 19, 2000

    I keep reading (over and over and over) about the ability to intercept information over the wire as it travels the Internet. This article states:

    But what if the scientific documents, as they travel across the Internet in an email message, fall into the hands of a competitor?

    However, I have never read a plausible scenario by which someone could actually do this. Assume I use Pacific Bell as my ISP. Theoretically, if someone were to have physical access to a PacBell facility, they could put a packet sniffer next to an Internet router and intercept my super-secret e-mail. I have a difficult time believing that breaking into a facility and physically planting such an item is that easy. OK, so suppose a machine was on the same physical device (hub or switch) as an Internet router. Theoretically, someone could break into that machine, upload a sniffer to it and intercept my e-mail. Again, I have a problem with this. Wouldn't the machine be on a switch rather than a hub? And if this were the case, wouldn't the switch defeat a promiscuous-mode sniffer? (Maybe I'm wrong here). And why would a major ISP put a machine that could potentially be broken into on the same physical device as an Internet router? OK, so suppose they've compromised a machine on my network and installed a sniffer on that. THEN they could intercept my super-secret e-mail. But they've already compromised my network, so what's the point?

    I am NOT a security know-it-all - I would call myself somewhat knowledgeable. I agree with the basic idea of the article - encrypting the traffic is better than not encrypting it. However, according to what I have read, Internet-connected hosts are broken into on a daily basis. I have never heard of someone sniffing anything off the wire, yet this method of information theft seems to be the focus of many security articles.

    I'm not saying that anyone is wrong and I'm not saying stealing information this way is impossible. What I'd like is for authors to educate me - I'd like them tell me HOW this information can be snatched off the wire - not just that it can be done.

You must log on before posting a comment.

Are you a new visitor? Register Here

advertisement

advertisement

Windows is a trademark of the Microsoft group of companies. Windows IT Pro is used by Penton Media Inc. under license from owner.