Multi-Scanning

Research shows that as more anti-malware engines are added, malware detection rates improve, since each engine may not detect certain types of threats. Each individual engine specializes in different categories. Since each anti-malware engine uses different algorithms, malware analysts are in different time zones and different geographically-based labs, the value of combining multiple anti-malware engines significantly increases detection.

As shown in our multi-scanning test of more than 10,000 of the most active threats, we achieved over 95 percent detection with 12 combined engines, over 97 percent detection with 16 engines, and over 99 percent detection with 20 or more engines. 

During malware outbreaks the time it takes to detect a new threat is critical. Even small changes in detection rates can add days, weeks or months to the time it takes for various anti-malware engines to respond to emerging threats.

AV-Test.org conducted a test that shows that the detection mechanisms used by different anti-malware engines are faster at detecting certain malware compared to others. By combining the results of multiple scanning engines, we can reduce outbreak exposure times and achieve virtually zero exposure.

These gaps in detection are cause for concern because they expose organizations that use only a single anti-malware engine that hasn't yet detected a specific threat. For example, the Nemucod.KP trojan was initially detected by three anti-malware engines on March 16th, 2016. Within two days eleven anti-malware engines had detected the threat, and after one week sixteen engines had detected it. But months later, 24 engines still had not detected the Nemucod.KP threat.

The ability for multi-scanning to dramatically reduce the exposure gaps of using one or a small number of anti-malware engines makes it a valuable approach for the early detection of emerging and actual outbreaks.

With multi-scanning, you can avoid exposures caused by the potential limitations of a single vendor. This could be a technology issue, like a particular vendor being unable to detect a vulnerability because of a technical limitation, or it could be a business reason, like a vendor not being allowed to operate in certain geographic regions or government agencies. For example:

• Symantec vulnerability
• Kaspersky U.S. government ban

Over-reliance on a single vendor can prove challenging, but these issues are avoided with multi-scanning approaches. Multi-scanning also gives you the flexibility of removing a problematic vendor from your deployment environment if vendor issues occur.

False positives, where files are reported as malicious when they are not, surface as a side-effect of any malware scanning solution, and can adversely affect business operations. To further complicate the issue, false positives are often only reported by a few anti-malware vendors at a time, and they are not always consistent or reproducible during testing.

False positive rates are reduced because many malware vendors work together through malware data sharing programs. This means that vendors work together to help codify true positives and false positives, so that overlapping vendor data has many fewer false positives, thus improving the results of using multi-scanning.

Also vendors share whitelist (trusted file) data. Our whitelist database accumulates the data from many vendors, which also reduces false positive detection rates.

Every engine returns some false positives, but it is incorrect to assume that using two engines results in double the number of false positives. Overlap in the detection of false positives using multi-scanning, limits the number of new false positives added by each new engine, as our multi-scanning research demonstrates. When we use more engines, the amount of false positives does go up, but only by a small, fractional amount, which is outweighed by the many benefits of multi-scanning.

Scanning with multiple engines takes slightly longer than scanning with a single engine, but with our multi-scanning methods, performance loss is minimized. Our methods take into account redundant tasks such as opening archives and detecting file types, and we also leverage the fact that various engines specialize in detecting threats in specific file types.  This means that many multi-scanning tasks can be parallelized by using methods like distributed computing, multi-core processing and scanning in memory.

Because multi-scanning requires multiple anti-malware engines from various vendors, cost is a factor. However, we partner with vendors to deliver optimized multi-scanning engine package options to provide beneficial Total Cost of Ownership (TCO) over time. By serving as a single point of contact, we reduce complexity in multiple scanning deployments for our global client base of government entities and organizations in virtually every industry including other security firms, aerospace and defense, healthcare services, critical infrastructure, and supply chain manufacturing.