Automation is being hailed as a way to take some of the heavy lifting away from overworked security operations teams. Security vendors are integrating automation into their point solutions to automate tasks such as security policy orchestration, change and configuration management, incident response playbooks, and other labor-intensive tasks.
This is a good start toward solving some of the challenges of managing the modern security stack. But we need to think more strategically about automation if we're truly going to solve cybersecurity workforce challenges and gain any kind of edge over hackers.
Most automation takes place at the front end of the cycle: the detection and prioritization of security alerts. A combination of threat intelligence feeds, SIEMs, and incident response platforms generate event and incident data and perform some level of automation (correlation, orchestration, change management, etc.). This automation is helpful, but I hear, on average, from security teams that they are only spending about 30% of their time on the front end of the cycle. It's what happens after a threat is detected, prioritized, and sent to the operations team that the real work begins.
In most organizations I've worked with, I see an estimated 40% of a team's resources being poured into manual investigation of incidents. This is often the most painstaking, lengthy part of the security life cycle. Analysts tasked with investigating and remediating security alerts often see more than 1,000 alerts per week from the more than 40 vendors deployed throughout their complex environment. The introduction of threat intelligence compounds this problem, as a single feed can generate more than 3.5 million indicators per month. Given the volume of data that must be evaluated and investigated, the average enterprise is ultimately throwing away more than 90% of its security data.
The remaining 30% of their time is focused on mitigation and reporting of the incident. These last two steps are the most important for learning from an incident and being better prepared for a future incident — yet most teams simply do not have the time or infrastructure to properly follow through on them. Once the lengthy investigation process is concluded, the results of that investigation are retained as independent, isolated reports. The technical details of the security incident are not stored or structured in a way that allows for automated correlations and are often missing the organizational context. Even when enterprises are creating their own indicators, they are manually maintaining lists of malicious IPs or domains in spreadsheets or text files rather than feeding those insights back into the system to be applied to future threats.
It's not enough to simply introduce automation. In order to extract the most benefit from automation, we need to holistically review incident response processes to find the areas where security analysts can engage in more critical thought and problem-solving.
Part of that means finding ways to automate the actual intelligence and do more of the analytical work in order to allow analysts to make quicker, more informed decisions. Beyond automating process elements, look for ways to automate correlation rules, historical analysis and coordinated communication between security devices. Intelligence automation will bring incident response to the next level.