The first step in designing the log monitoring system is to evaluate the PCI requirements and determine which ones would be in scope. The detailed requirements list from the PCI DSS 1.1 is given in the sidebar. The requirements that might be satisfied with a log monitoring system include:
10.1 – Logging is enabled and active – We need to incorporate these sub-requirements into our log monitoring system design.
10.4 – Time Synchronization – We would need to configure the log monitoring servers to use the same NTP source as the log clients.
10.5 – Audit trail security – We need to incorporate these sub-requirements into our log monitoring system design.
10.6 – Log Review – We can incorporate these sub-requirements into our log monitoring system design.
10.7 – Log Retention – We can incorporate these sub-requirements into our log monitoring system design.
Note that 10.2 and 10.3 would be out of scope for the log monitoring system as these specify the types of events and the metadata that would be recorded. Obviously log content and generation would need to be configured appropriately at each monitored device.
Now that we know what we’re trying to achieve, let’s examine the possible design concepts and ways in which we can meet the requirements. Perhaps the best way to start is by including a centralized log server in the design. Although only sub-requirements 10.5.3 and 10.5.4 specifically calls for a log server, we’ll see that using the centralized design will allow us to be more flexible in meeting the other requirements and will also only require minimal changes to the production servers in the PCI network.
If you have a geographically distributed network, then you should take the quantity of log traffic into account. Assuming the default syslog protocol is used (as opposed to newer, more enhanced protocols such as syslog-ng) the log messages will be UDP based and could be discarded when WAN traffic is heavy. Therefore, you will probably want to distribute your log monitoring system with at least one log server per datacenter. This will reduce the amount of WAN traffic to a minimum. Don’t forget to configure your wireless APs to communicate with the centralized log server in order to satisfy requirement 10.5.4!
Once we have the logs collected in a centralized location, we can start applying additional controls around the logs that satisfy the other PCI sub-requirements. Let’s start with 10.6, which requires us to ensure that periodic log reviews are performed. The revised version 1.1 DSS explicitly allows for the use log monitoring/alerting tools to achieve compliance. There are many commercial and some open source tools available that we can implement to handle this monitoring and alerting. Since we’ll be running this process only on the centralized log server itself we also won’t have any performance hits to the rest of the production servers on the PCI network.
Once our log alerting is in place, meeting requirement 10.1 will be quite simple. We can state that any server from which we have received log data in the past arbitrary time period is considered to have its logged capabilities enabled and active. Should a particular server not check in within the arbitrary time period, then we would want to generate an alert for that server.