9.2. Host-based IDS

A host-based IDS analyzes several areas to determine misuse (malicious or abusive activity inside the network) or intrusion (breaches from the outside). Host-based IDSes consult several types of log files (kernel, system, server, network, firewall, and more), and compare the logs against an internal database of common signatures for known attacks. UNIX and Linux host-based IDSes make heavy use of syslog and its ability to separate logged events by their severity (for example, minor printer messages versus major kernel warnings). The syslog command is available when installing the sysklogd package, which is included with Red Hat Enterprise Linux. This package provides system logging and kernel message trapping. The host-based IDS filters logs (which, in the case of some network and kernel event logs, can be quite verbose), analyzes them, re-tags the anomalous messages with its own system of severity rating, and collects them in its own specialized log for administrator analysis.

A host-based IDS can also verify the data integrity of important files and executables. It checks a database of sensitive files (and any files added by the administrator) and creates a checksum of each file with a message-file digest utility such as md5sum (128-bit algorithm) or sha1sum (160-bit algorithm). The host-based IDS then stores the sums in a plain text file and periodically compares the file checksums against the values in the text file. If any of the file checksums do not match, the IDS alerts the administrator by email or cellular pager. This is the process used by Tripwire, which is discussed in Section 9.2.1, “Tripwire”.