Server virtualization will become a dominant factor in the next three
to five years in an effort to reduce operating costs and simplify
business. One highly popular trend is using virtualization for data
center consolidation. Companies are consolidating everything from Web
servers to the servers that run CRM applications, all in an effort to
streamline operations and create efficiency across-the-board.
The consolidation of physical assets to form a heterogeneous environment, while significantly reducing overhead, sounds like
a very attractive proposal for any CIO. However, it also introduces
additional problems that can substantially increase security risks. Simply put the aggregation of multiple functions and resources into a
single physical platform will not only increase your overall risk, but
introduce a single point of failure. This holds especially true if the
system has functions relating to the storage and retrieval of sensitive
Decisions to adopt virtualization aren’t primarily driven from a
security viewpoint, but from business enablement. Thus, administrators
may not fully understand the risks and implications associated with the
deployment of virtualization.This article will look at several key issues related to the evolving
threat landscape and offer advice on how to mitigate these threats.
Virtualization Security: It Starts with Hyperjacking
When looking at the risks inherent in virtualization, we need to
understand a bit about the basic architecture – starting with the
hypervisor that represents the primary abstraction layer between the
physical hardware and the virtual machines (VMs) that are running.
While being able to consolidate resources saves money, it introduces
the possibility of “hyperjacking” – malicious control of the hypervisor. The hypervisor represents a single point of failure when it comes to
the security and protection of sensitive information. Theoretically, if
this layer is compromised, all the VMs that are running could be
accessed by the bad guys. This substantially increases exposure,
because it gives hackers another access point to a company’s internal
But if a hacker compromises a Web server connected to a customer
back-end database via a SQL injection, only the information at that
source is compromised, reducing the overall impact of the breach.
However, virtualization by its nature puts multiple servers and
multiple data sources at risk, substantially increasing the degree of
For example, if a company has a clustered group of SQL servers
running on the same hypervisor, hackers could target and compromise
that layer through a number of different ways, such as subverting the
hypervisor with targeted malware.
The virtual machine is essentially a file or an image stored on a
hard disk. It provides the network administrator with the flexibility
and control to move a virtual system from one physical platform to
another. And there are tools to assist in migrating active VMs to other
live physical servers without interruption. However, because the
machine exists as a file, it’s subject to attack by viruses and other
malware designed to infect the associated file formats (e.g., VMDK,
Furthermore, the VM can be accessed offline by remounting the image,
allowing the hacker to gain access to the applications and the data
stored in them. Offline brute-force dictionary attacks are now possible
thanks to the portability that virtualization offers.
Loss of Visibility
Because virtualization fundamentally changes the way that
infrastructure is deployed, losses in visibility can occur since the
resources are consolidated into a single system. This increases
complexity in terms of accurately logging security events.
Tools that were once designed for monitoring access to physical
servers now have to be redeployed and reconfigured to monitor multiple
aspects of the system, not just the host operating system but the VMs
running on top. It’s important to rethink your strategy to include methods for monitoring access to resources.
Strategies for Mitigating Your Risk
When considering a strategy to minimize the risks associated with
virtualization, a holistic approach is best – meaning multiple
methodologies should be used to thwart hyperjacking.
Because we are dealing with machines within machines, we have to pay
special attention to protecting the virtual machine as well as the core
architecture – essentially the host operating system running the
hypervisor. In this context, further attacks are likely to come from
within. In other words, hackers may attempt to subvert the hypervisor
to inject targeted malware in an effort to gain access to the VMs.
Fortunately, several different technologies exist that
administrators can use to implement a strategy to proactively thwart
possible attacks. They include:
- Database Monitoring: Technologies exist to monitor SQL and
Oracle databases for suspicious activity (access from unauthorized
users, script insertion, SQL statement execution, etc.). Monitoring is
only part of the equation in detecting an actual breach in progress. If
hackers subsequently decide to access the information stored in your
databases, besides extracting the data in real-time, database
monitoring will increase the odds of discovering unauthorized access.
- Network Intrusion Detection: Intrusion detection
technologies, in addition to other methods, can be used to detect
anomalous traffic and behavior that might be associated with an attack.
- Hardening Critical Assets: You can minimize your exposure
and risk by hardening critical assets (in this case the host system).
In other words, you remove non-essential functionality such as
services, applications, and ports that both add to the complexity, but
introduce additional risk.
It’s important that we take a proactive approach when developing a
security plan for a virtual network because the rate at which new
malware appears outweighs the capabilities of anti-malware labs to keep
up with new threats. The best security policy should include preventive
strategies designed for mitigating threats to virtualization.