Will Technology Replace Security Analysts?
Posted on September 15, 2014 by Kara Dunlap in Security
Recently, at a round table discussion, I heard someone make the statement, “In five years, there will be no more security analysts. They will be replaced by technology.” This is not the first time I have heard a statement along these lines. I suppose that these sorts of statements are attention grabbing and headline worthy, but I think they are a bit naïve to say the least.
Taking a step back, it seems to me that this statement is based on the belief or assumption that operational work being performed within the known threat landscape of today can be fully automated within five years. I don’t know enough about the specific technologies that would be involved in that endeavor to comment on whether or not that is an errant belief. However, I can make two observations, based on my experience, which I believe are relevant to this specific discussion:
• Operational work tends to focus on known knowns
• The threat landscape of today, both known and unknown, will not be the threat landscape of tomorrow
The work that is typically performed in a security operations setting follows the incident response process of: Detection, Analysis, Containment, Remediation, Recovery, and Lessons Learned. Detection is what kicks off this process and what drives the day-to-day workflow in a security operations environment. If we think about it, in order to detect something, we have to know about it. It doesn’t matter if we learn of it via third party notification, signature-based detection, anomaly detection, or any other means.
The bottom line is that if we become aware of something, it is by definition “known”. But what percentage of suspicious or malicious activity that may be present within our organizations do we realistically think is known? I don’t know of a good way to measure this, since it involves a fair amount of information that is unknowable. I do, however, think we would be naïve to think it is anywhere near 100%.
If we take a step back, the ramifications of this are quite striking. In essence, most of the work we are performing today involves what is likely a mere fraction of what ought to concern us. Even if technology could automate all of today’s security operations functions within five years’ time, that still leaves quite a bit of work undone.
I think we would also be naïve to think that the threats of today, both known and unknown will be the threats of tomorrow. If I think back five or ten years, I’m not sure how many of us foresaw the degree to which intrusions involving large-scale payment card theft would become an almost regular occurrence. Granted, theft of sensitive information has been an issue for quite some time, but not to the degree that it has been in the recent past. Payment card theft is now a threat that many organizations take very seriously, whereas five or ten years ago, it may have been a threat that only certain specific organizations would have taken seriously. This is merely an example, but my main point here is that we can’t view today’s threat landscape as a base upon which to build predictions and make assertions for the future.
In my experience, analysts can provide unique value that may not be obvious to those who have not worked in the role. For those who don’t know, I worked as an analyst for many years before moving over to the vendor side. It is from that experience that I make this point.
In a mature security operations setting, there will be a certain workflow and process. Some organizations will be in a more mature place, while other organizations will be in a less mature place. Regardless of where an organization finds itself, there will always be room to improve. Alongside performing the tasks required by the process, a good analyst will make the process better and improve the maturity of the organization. This can happen in many ways, but here are a few different approaches that I have often seen:
• Improving existing alerting
• Identifying automation opportunities
• Performing gap analysis
• Implementing new alerting
Any work queue will have both signal (true positives) and noise (false positives). A mature, efficient security operations program will have a high enough signal-to-noise ratio so as to allow for a reasonable chance at timely detection of incidents. Regardless of the signal-to-noise ratio, alerting that populates the work queue can always be improved. As the people most familiar with the ins and outs of various different alerts, analysts play an important role here. The analyst can provide unique perspective regarding tuning and improving alerts to make them less noisy and ensure they keep up with the times.
It is certainly true that some alerts follow a nearly identical sequence of events each time they are vetted, qualified, and investigated. These cases are good candidates for automation, but they won’t identify themselves. A skilled analyst is needed to identify those repetitive manual tasks best suited for automation. Automation is a good thing and should be leveraged whenever appropriate, but it will never replace the analyst.
With automation comes newly liberated analyst cycles. Those cycles can and should be used to hunt, dig, and perform gap analysis. Hunting and digging help to identify unknown unknowns – network traffic or endpoint activity for which the true nature is unknown. Gap analysis serves to identify points within the organization where proper network and endpoint telemetry may not exist. All these activities help provide a window into the unknown. After all, today’s unknown may be tomorrow’s breach.
When unknown unknowns are discovered, they should be studied to understand their true nature. This process turns them into new known knowns. And it is from this pile that new alerting is continuously formulated. The analyst is an invaluable resource in turning unknown unknowns into known knowns. Based on my experience, there is no shortage of unknown unknowns waiting to be investigated.
A good analyst is hard to find and is a critical resource within a mature security operations function. Although it may be tempting to envision a world where the analyst has been fully automated, this does not seem particularly reasonable. Rather, the work of the analyst can and must evolve over time to keep pace with the changing threat landscape.
Joshua Goldfarb (Twitter: @ananalytical) is Chief Security Strategist of the Enterprise Forensics Group at FireEye and has over a decade of experience building, operating, and running Security Operations Centers (SOCs). Before joining nPulse Technologies, which was acquired by FireEye, as its Chief Security Officer (CSO), he worked as an independent consultant where consulted and advised numerous clients in both the public and private sectors at strategic and tactical levels. Earlier in his career Goldfarb served as the Chief of Analysis for US-CERT where he built from the ground up and subsequently ran the network, physical media and malware analysis/forensics capabilities. Goldfarb holds both a B.A. in Physics and a M.Eng. in Operations Research and Information Engineering from Cornell University.
Tags:
