In a previous blog post, I outlined the threat of attack by exploitation of weak links— attack vectors innocuous enough to appear on the radar, but exploitable enough to move an attacker one step further into your organization.

Using lateral movement to penetrate an organization is like playing a board game. You don’t leap across the board in one swoop. Instead, you move square by square (hopefully without your opponent noticing you) until you can strike.

So, when it comes to defending your organization against threats, you are not playing one game board against one opponent. You are playing against multiple opponents on hundreds or thousands of game boards (your systems) simultaneously. So just how do you keep track of these moves? And, how do you distinguish a valid users’ activity from an attacker’s?

Anomaly Detection and Threat Analytics

Logging user activity is a best practice for most organizations, but often you get so much information back that it is very difficult to manually identify patterns of behavior in audit trails. This is where threat analytics comes to the rescue. Self-learning algorithms learn what activity standard behavior is made of (this is also known as baselining.) Whenever something changes, it can generate an event to alert you to the fact that a user has done something different.

So, we can just implement a baselining tool and we’re covered, right?

Not quite. Patterns of behavior can be quite complex, and often it’s not a single anomaly that stands out— it’s a collection of anomalies threaded together. I’ll give an example:

Let’s say I have a rule that highlights when a user logs onto a new server; this rule may well be firing off hundreds of times a day as administrators build up and tear down virtualized environments. We’ll simply give this anomalous behavior a score of “1”. Now we look at what time of day the user is making the access request— we’ll score all those events as “1”. What about servers that have unpatched vulnerabilities? I think that access to these systems are a little riskier, so we’ll give these events a score of “1”.

What you will see is that by looking at activity from multiple angles and incrementing up the activities score based on different factors, we can start to see events that “pop out”. To give another example:

A user tried to access a vulnerable server they had not used before, at a time they do not normally work. The access request came over a VPN from a foreign location, and the user tried to jump to another host using the MSTSC command. These factors combined to elevate the event threat level enough that it stood out and received immediate attention.

BeyondTrust’s advanced threat analytics solution helps pinpoint breaches that many other products miss. It processes and analyses data from network devices such as Palo Alto, as well as BeyondTrust’s full range of Privilege Access Management and Vulnerability management solutions, including:

By processing activity coming from the vulnerability and delegation solutions, as well as network devices, BeyondTrust’s threat analytics solution can add rich context to any single event whether host or network initiated.

If you’d like to speak with us to learn how we can help you identify and eliminate potential threats in your organization, contact us today.

Up next

From September 27, 2018:

In Cyber-warfare, Speed Heals