Malicious Code Propagation and Antivirus Software UpdatesRelease Date: July 2, 2003
Recent reports to the CERT/CC have highlighted two chronic problems:
- The speed at which viruses are spreading is increasing. This
echoes the trend toward faster propagation rates seen in the past few
years in self-propagating malicious code (i.e., worms). Beginning
with the Code Red worm (CA-2001-19, CA-2001-23) in 2001 up through
the Slammer worm (CA-2003-04) earlier this year,
we have seen worm propagation times drop from hours to minutes.
A similar trend from weeks to hours has emerged in the virus (i.e., non-self-propagating malicious code) arena. The effectiveness of antivirus software suffers as a result. Several recent malicious code incidents involving variants of W32/BugBear and W32/Sobig have achieved widespread propagation at rates significantly faster than many previous viruses. This increased speed is, unfortunately, also faster than many antivirus signatures can be identified and updated, regardless of the update method (including automated signature updates). The CERT/CC has received reports of successful W32/Sobig.E compromises from users whose signatures were up to date for the prior versions of W32/Sobig.
Signature-based antivirus software is not the only type of antivirus software at risk: antivirus software that uses heuristics to determine malicious behavior may be circumvented by malicious code that employ new techniques. They should not be unconditionally trusted either, as they may not always block malicious code from executing. Additionally, we are aware of instances where corrupted antivirus software updates have caused the software to be disabled without the user's knowledge.
- In a number of the reports, users who were compromised may have been under the incorrect impression that merely having antivirus software installed was enough to protect them from all malicious code attacks. This is simply a mistaken assumption, and users must always exercise caution when handling email attachments or other code or data from untrustworthy sources.
In general, it is important to remember that while antivirus software vendors continue to improve the speed and reliability of their signature update mechanisms, there will always be some window of time when a system does not contain signatures to detect a particular worm or virus. Several recent research papers that have placed estimates on the magnitude of "worst-case scenario" malicious code propagation rates also illustrate the risk to systems during the window of time before signatures are available.
Apply "defense in-depth"
As mentioned above, it is not sufficient to rely solely on antivirus software for complete protection. Therefore, we recommend users apply a strategy of "defense in-depth" (where several layers of security or access controls are used) when considering ways to protect their computers from attackers. Although it may not be practical for all users, another way of achieving defense in-depth is to use diverse software and operating systems when possible. Some additional ways of improving security beyond the use of antivirus software follow.
Run and maintain an antivirus product
While an up-to-date antivirus software package cannot protect against all malicious code, for most users it remains the best first-line of defense against malicious code attacks.
Most antivirus software vendors release frequently updated information, tools, or virus databases to help detect and recover from malicious code, including W32/Bugbear.B and W32/Sobig.E. Therefore, it is important that users keep their antivirus software up to date. The CERT/CC maintains a partial list of antivirus vendors.
Many antivirus packages support automatic updates of virus definitions. The CERT/CC recommends using these automatic updates when available.
Do not run programs of unknown origin
Never download, install, or run a program unless you know it to be authored by a person or company that you trust. Email users should be wary of unexpected attachments, while users of Internet Relay Chat (IRC), Instant Messaging (IM), and file-sharing services should be particularly wary of following links or running software sent to them by other users, as these are commonly used methods among intruders attempting to build networks of distributed denial-of-service (DDoS) agents.
Disable or secure file shares
Best practice dictates a policy of least privilege. For example, if a Windows computer is not intended to be a server (i.e., share files or printers with others), "File and Printer Sharing for Microsoft Networks" should be disabled.
For computers that export shares, ensure that user authentication is required and that each account has a well-chosen password. Furthermore, consider using a firewall to control which computer can access these shares.
By default, Windows NT, 2000, and XP create certain hidden and administrative shares. See the HOW TO: Create and Delete Hidden or Administrative Shares on Client Computers for further guidelines on managing these shares.
Deploy a firewall
The CERT/CC also recommends using a firewall product, such as a network appliance or a personal firewall software package. In some situations, these products may be able to alert users to the fact that their machine has been compromised. Furthermore, they have the ability to block intruders from accessing backdoors over the network. However, no firewall can detect or stop all attacks, so it is important to continue to follow safe computing practices.
Recovering from a system compromise
If you believe a system under your administrative control has been compromised, please follow the steps outlined in
- Paxson, V., Staniford, S., Weaver, N. "How to 0wn the Internet in Your Spare Time" http://www.icir.org/vern/papers/cdc-usenix-sec02/index.html
- Moore, D., Paxson, V., Savage, S., Shannon, S., Staniford, S., Weaver, N. "The Spread of the Sapphire/Slammer Worm" http://www.cs.berkeley.edu/~nweaver/sapphire/
Authors: Chad Dougherty and Allen Householder
Copyright ©2003 Carnegie Mellon University.