Rollout: How Much Is Bot Detection Worth To You?

Damballa's appliance shows promise, but it still has a lot of ground to cover.
Gone are the days when a malware-infected system would display a silly image or corrupt itself just to annoy the user. Now we face bot armies that check in with their handlers for their next task, which could be a distributed denial-of-service attack against government systems or to capture and transmit sensitive data such as bank account information or encryption keys back to the bot owner. Cybercrime has become big business, and botnets are part of that business.

A group of appliance vendors, including Damballa, is rising to meet the challenge. Damballa's Failsafe appliance/software package aims to put bots out of business, or at least dent their capabilities. Tests showed Damballa's approach would be a good fit for large enterprises with other malware protections already in place, but it might not be the best bet for organizations that need to stretch the security budget or for those deploying anti-malware systems for the first time.

Founded in 2006 with technology developed at Georgia Tech, Damballa focuses on protecting businesses from targeted attacks. Utilizing software, global monitoring, and traffic analysis, Damballa claims its Failsafe appliance can detect malicious botnets within organizations even when those organizations have up-to-date antivirus and malware protections.

According to Damballa, 3% to 5% of enterprise network assets worldwide are infected with bot malware. By not relying on signatures like intrusion-detection and -prevention systems, Damballa claims it can detect and report on these and other zero-day malware through real-time traffic analysis.

The Failsafe v3.0 appliance is a custom-built server with four hot-swap hard drives, redundant power supplies, five network interfaces, and enough power to monitor at a minimum 10,000 nodes. The appliance is licensed per node, with a minimum purchase of 10,000 nodes priced at $100,000.

Installation is straightforward: The appliance uses a span port or network tap to see network traffic. Since the Failsafe appliance does not sit in-line, it's not a point of failure in your network architecture. VPN connectivity from the appliance to Damballa is required for remote support, upgrades, and uploading of malware samples. The server itself is a standard rack-mount unit.

Our Take
At $100,000 or more, Damballa Failsafe 3.0's bot-only focus will keep it out of organizations that need comprehensive threat protection.
Despite its 3.0 version number, Damballa is a "young" product, and it shows in limited capabilities.
Damballa detects bots and warns admins, but can't stop the trouble it's found.
Once connected and racked, setup is performed through a console-based menu system that offered no unpleasant surprises in tests. After setting our management IP address and configuring a host name and other associated network settings, we were ready to log in through the Web-based management console.

Once you log in to the Web-based console, you'll see just how straightforward -- and early stage -- Failsafe really is. You don't need to do much, and there isn't much for you to do: You can change some account information, look at reports, or read help documents, and that's about all. To lose its rookie status, the Failsafe appliance must offer staff more useful features, such as the ability to remotely remediate infected systems from the appliance or provide more in-depth reporting such as the ability to report on infections by user, not just system.

Tests In Traffic
To ensure that the appliance was monitoring our traffic, we began with some simple activities. The Damballa appliance can track the number of binary files downloaded, so we thought this would be a good test of monitoring ability. We downloaded several binary files, including the Windows-based Putty SSH client and the Windows-based uTorrent BitTorrent client. Both of these files were marked as malware and reported back to Damballa. This proved the appliance was in fact monitoring our traffic, but also that there's room for improvement in the analysis of binaries.

Next, we decided to test real command-and-control traffic. The appliance was moved to one of our isolated test networks where we sneaked infected bot files onto our test hosts through a back channel to ensure that the appliance could not see them. Once loaded, we let the malware connect to its handlers to receive commands. The appliance was able to detect this activity and report the host that was compromised. If this were the real world, our IT staff could now be dispatched to clean the infected host and perform a postmortem.

Unfortunately, that's all that can be done. As of Failsafe 3.0, there are no threat-blocking, mitigation, or remediation features. Once a system is found to be infected, staff should race to clean it and hope other controls have stopped sensitive information from leaving the company. Damballa is quick to point out that a proper security posture is made up more than one single control, and its approach seems to be to rely on other controls to stop the threat.

At $100,000 for the privilege of monitoring 10,000 nodes, the product is lacking in the return-on-investment department when compared with offerings from Blue Coat Systems, McAfee, Mi5 Networks, Symantec, and others.

Damballa is hyperfocused on the threat of bots, and thus in theory should be able to detect these threats faster than other products. This may be worth a premium to some organizations, but for others, a more common approach to virus, malware, and bot detection and prevention may be a better fit -- especially those that can't afford the price of Damballa plus other protection systems.

It's worth noting that Damballa realizes the ability to block, whether automated or manual, and host cleaning are features worth pursuing. This product works as advertised, but has a high price tag and is in its early stages.

Adam Ely is senior manager of technology at a Fortune 100 company and a frequent contributor to InformationWeek.