The incessant escalation, both in malware sophistication and proliferation, means the need for fundamental file integrity monitoring is essential to maintain malware-free systems. Signature-based anti-virus technologies are too fallible and easily circumnavigated by zero-day malware or selectively created and targeted advanced persistent threat (APT) virus, worm or Trojan malware.
Any good security policy will recommend the use of regular file integrity checks on system and configuration files and best practice-based security standards such as the PCI DSS (Requirement 11.5), NERC CIP (System Security R15-R19), Department of Defense Information Assurance (IA) Implementation (DODI 8500.2), Sarbanes-Oxley (Section 404), FISMA - Federal Information Security Management Act (NIST SP800-53 Rev3) specifically mandate the need to perform regular checks for any unauthorized modification of critical system files, configuration files, or content files; and configure the software to perform critical file comparisons at least weekly.
However, file-integrity monitoring needs to be deployed with a little advanced planning and understanding of how the file systems of your servers behave on a routine basis in order to determine what unusual and therefore potentially threatening events look like.
The next question is then whether an Agentless or Agent-based approach is best for your environment. This article looks at the pros and cons of both options.
Agentless FIM for Windows and Linux/Unix Servers
Starting with the most obvious advantage, the first clear benefit of an Agentless approach to file integrity monitoring is that it doesn't need any agent software to be deployed on the monitored host. This means that an Agentless FIM solution like Tripwire® or nCircle will always be the quickest option to deploy and to get results from. Not only that but there is no agent software to update or potentially interfere with the server operation.
The typical Agentless file-integrity monitoring solution for Windows and Linux/Unix will utilize a scripted, command-line interaction with the host to interrogate the salient files. At the simplest end of the scale, Linux files can be baselined using a cat command and a comparison done with the subsequent samples to detect any changes. Alternatively, if a vulnerability audit is being performed in order to harden the server configuration, then a series of grep commands, used with regex expressions, will more precisely identify missing or incorrect configuration settings. Similarly, a Windows server can be interrogated using command line programs, for example, the net.exe program can be used to expose the user accounts on a system, or even assess the state or other attribute associated with a user account if piped with a find command e.g. net.exe users guest |find.exe /i "Account active" will return an "Account active Yes" or "Account active No" result and establish if the Guest account is enabled, a classic vulnerability for any Windows server.
Agent-Based File Integrity Monitoring
The key advantage of an Agent for FIM is that it can monitor file changes in real-time. Due to the agent being installed on the monitored host, the OS activity can be monitored and any file activity can be observed and changes recorded. Clearly any Agentless approach will need to be operated on a scheduled poll basis and inevitably there will be a pay-off between the frequency of polls being regular enough to catch changes as they happen, and the limiting the increased load on the host and network due to the monitoring. In practice polling is typically run once per day on most FIM solutions, for example Tripwire®, and this means that you risk being anything up to 24 hours late to identify potential security incidents.
The second major advantage of an agent-based file-integrity solution is that the host does not need to be 'opened up' to allow monitoring. All critical system and configuration files will always be protected by the host filesystem security, for example, the Windows System32 folder is always an 'Administrator Access Only' folder. In order to monitor the files in this location, any external scripted interaction will need to be provided with Admin rights over the Host and this immediately means that the host needs to be made accessible via the network and an additional User or Service Account needs to be provisioned with Admin privilege, potentially introducing a new security weakness to the system. By contrast, an Agent operates within the confines of the Host, just pushing out File Integrity changes as they are detected.
Finally having an Agent offers a distinct advantage over and above the Agentless approach in that it can offer a 'changes only' update across the network, and even then only when there is a change to report. The Agentless solution will need to run through its complete checklist of queries in order to make any assessment of whether changes have been identified and even using elaborate WMI or Powershell scripts still requires considerable resource usage on the host and the network when dragging results back.
Nobody likes installing and maintaining agents on their servers and, if this can be avoided, this is an attractive option to take.
However, by not using an agent, the Agentless interrogation of the host to gather file integrity monitoring results will inevitably end up being more resource-intensive in terms of server cpu and memory, not to mention network bandwidth. Furthermore it can be argued that a truly real-time FIM solution must be agent-based because Agentless means scheduled polls and inevitably there will be gaps between polls.
So in summary, the only solution is to use a FIM solution that supports both agentless and agent-based FIM in order to give you complete freedom and flexibility for all platforms and scenarios.