There are quite a few architectures running around for Cloud and virtual environments, but except for a few, they seem to all be missing the ability to gain access to Application Performance Management (APM) data as a means to provide an early warning system for security issues.
Most security reference architectures rely on the old methods to get warnings about security issues such as use of a SIEM and a log analysis tool to interpret what is in the SIEM. However, there is a richer set of more immediate data that can help us with the problem of security notifications: APM Data.
APM Data provides a rich and different approach to security early warnings but the interpretation of the APM Data implies knowledge of the application that security professionals may not have. Yes, this is not a requirement as the security team and the applications team will be solving problems together that come up when there is an anomaly within any APM Data. The application team wants to know why there is an anomaly, perhaps a code path was taken unexpectedly, while the security team wants to insure that code path was not a hack attempt.
There are several ways to do this:
- Application and security professionals working together to determine if the APM Data shows a security issues or a code issue
- APM tools with built in mechanisms that could be used for security, such as a list of websites from which data comes into the system and to which data flows out of the system.
- APM tools that self learn the code path, so that when a new code path is used both security and application teams are notified
- APM tools that show both teams data about the code path when anomalies occur. Perhaps going so far as to highlight what was different
- APM Tools that show the exact process of events such as a database query to be investigated. Perhaps there was a SQL Injection within the query
APM tools have a rich set of data that could be used by security professionals. These tools know more about what is happening within an application than almost anyone else and could be helpful as a part of defense-in-depth. The smarter the APM tool, the more useful it becomes for security purposes.
Minimally, APM tools must contain the following abilities to be useful by security professionals:
- A way to see when external to the application resources were accessed, such as an external website.
- A way to see all database queries (even obfuscated if the APM solution is in the Cloud).
- A way to know when an anomaly has occurred, perhaps a different database query was made (possible SQL injection) or some normally unused code path was taken.
- A way to know when performance changes, perhaps activity is happening too fast (which could imply a DoS attack) or too slow (misconfigured or malware present).
In the end, however, it is all about determining when something anomalous has happened and a means of providing that data to the security team as well as the application team so that both work the problem side by side.
ABOUT Edward L. Halekty
Edward L. Halekty is Virtualization and Cloud Analyst, The Virtualization Practice LLC.