Postmortem: What to Do After a Security Incident

Incidents happen. Vulnerabilities happen. The quality of your response can make the difference between a bad day and a disaster. What happens after the response can make the difference between endless firefighting and becoming stronger with every battle. A quality postmortem analysis is free ammunition. Every incident is someone or some event showing where a system's weaknesses are, if only one is willing to listen.

This is how a good information security officer, or an engineer who's a true information security evangelist, can make a difference:

  1. Something happens. It may be an exercise, or a real incident.

  2. You now have real information to go on. You are in a very different position from when you were working from the theoretical.

  3. If you know how to understand that information, and what information you need, you may have a new understanding of the project or organization's security needs. Even if this is only confirmation of what you knew before, it is important because...

  4. This information and analysis, if communicated effectively, especially in the aftermath of an incident, can be a powerful tool for fixing problems.

  5. Next time around, the organization will be a little more on its game, and another set of weaknesses can be shored up. Every iteration makes the organization stronger.

How to Sabotage Your Postmortem

Postmortem mistakes can have long-term implications, but they also can take a long time to identify. A bad postmortem feels just as satisfying as a good postmortem to someone who doesn't know the difference. Unfortunately, it fills a team—or a whole organization—with false beliefs, missed opportunities and bad data, eroding its ability to mature its security. These erosions are small individually, but like water lapping up against a beach, they eventually aggregate. Learn these anti-patterns and be certain to recognize them.

Play the blame game.

Yes, some incidents are clearly one person's fault. Most of the time though, there's plenty of blame to go around. Blame is an out that makes it too easy to ignore systemic problems, and looking for someone to punish makes valuable sources of information go silent. Deal with personnel issues separately from incident postmortem, except in cases of actual malicious insider attacks.

Stop at the vulnerability.

Calling it quits once you've found something to patch or a configuration to change is perhaps the most common mistake. In the best of cases, looking deeper can confirm what's working and what isn't. In the majority of cases, there is more than one cause to be found. Don't stop looking once you've found something; poke in all the corners first.

Stop at the forensics.

Another common mistake is to look for signs of technological vulnerability or compromise, such as incorrect firewall configurations, software bugs, rootkits and so on without looking at the bigger picture of what may be causing those things to happen. Poor software engineering practice or inadequate tools for engineers will raise the incidence of bugs. So will overwork and poor morale. Similarly, a lack of configuration management for systems can cause mistakes due to forcing administrators to repeat processes by rote many times.