Interview with Richard Bejtlich -- GE Director of Incident Response
Back in May, I attended a meeting to get a feel for the company and group I would be working for this summer as an IT Security Intern. Much to my surprise, Richard Bejtlich was in attendance and as it turned out we'd be working for the same company. Anyways, Richard agreed to do be interviewed on network security monitoring and his new role as Director of Incident Response.
From reading your blog, most of us know you've served in the US Air Force. How has serving as an Air Force intelligence officer prepared you for a career in information security?
My Air Force intelligence training taught me to analyze data and search for patterns. It gave me some historical background on related fields like SIGINT. I also attended some specialized training, like the Defense Intelligence Agency Indications and Warning course. This provided me with a way to look at events as indicators that can be analyzed to form warnings, which are then escalated to decision makers. I also learned proper definitions for terms like "threat" and learned how to model threats.
What is the most important lesson you learned serving on the AFCERT?
I learned many lessons there. I wouldn't be where I am today if I had not joined the AFCERT. The first that comes to mind is the importance of collecting alert, session, statistical, and full content data. The AFCERT was practicing NSM before it really had a name for it. NSM was the name of the sensor built by Todd Heberlein. My entire detection and response methodology is built on that AFCERT foundation. I also learned to appreciate that people are the real threats, not malware or tools. Malware and tools change and disappear, but people rarely do, and people are the bad guys.
What made you want to start your own company, TaoSecurity?
While working at ManTech, many customers were approaching me directly for consulting and training help. I thought, "They are trying hire me, not ManTech. Why not remove the middle man?" So I started independent consulting in June 2005 and ended in June 2007.
Every company has ups and downs -- were there ever times you wanted to just quit [TaoSecurity] and what kept you chugging along at the end of the day?
Well, eventually I did quit. I decided I wanted to try working with a company for the long term, instead of working in one or two week bursts. Consulting exposes you to many customers, but if you want to really make a difference for the long term it pays to be an employee.
What makes a good network security analyst?
First, you need to want to beat the bad guys. If you are entering the security field because you heard a commercial on the radio advertising higher pay, you will not get far. You need to understand the business you are protecting, the processes and the technologies. I recommend having some system administration experience. I've had multiple students in my classes who do not have the foggiest notion where to look for the files that comprise a static HTML Web site, for example. You have to understand the attacks the adversary employs. I've talked with people who "play defense," but who have no interest whatsoever in learning how the offense operates. If you don't know offense, how are you going to play defense? I think it's important to read because it helps you stay current. You also need a curious mind and be detail-oriented so you can perform investigations.
Take me through a typical day with Richard Bejtlich, what do you typically do in the morning? Are you watching packet captures all day? Readers tend to see only the juicy stuff an NSM expert tackles by way of your blog, but how much of your time is spent with very mundane stuff?
My new job as Director of Incident Response is different from my previous work. During my first month at work I've spent a lot of time learning about the company and its detection and response requirements. This has been mostly non-technical, although I've been providing technical expertise for ongoing projects and cases. I'm working on assembling an enterprise sensor grid collecting data from various sources. When that begins to take shape I will have data to investigate, so I will return to that aspect of work.
I expect at some point to train analysts to do incident detection and response, so I will be left with the more complex material and the big picture issues. I try to spend a few minutes each day reviewing my Bloglines feeds to keep up with current security issues. I also try to read a few pages from a book on my reading list. If I get any real reading done, however, it's usually between 9 pm and 12:30 am.
What path would you recommend to those interested in entering the field of information security and more specifically, network security monitoring?
If you are in middle school, try to get into a tech-oriented high school. Then go to college for a CS degree. Get a job as a consultant or with a MSSP. Whatever tech job you have, you can integrate security.
You MUST have a home lab. You MUST run operating systems besides Windows. Knowledge of Windows is important because many of the victim systems you will investigate will be Windows. However, you can do far more with old hardware and zero budget when you open up to the open source world. I recommend trying FreeBSD. In fact, try a book like Building a Server with FreeBSD, and then read Absolute FreeBSD, 2nd Edition. I also recommend building your own sensor infrastructure to watch your lab. (I use Sguil.)
I also recommend reading my books, starting with Tao. I wrote it specifically for people just starting out. I got the idea to write it in 2001 when I was building a MSSP with Sguil developer Bamm Visscher. Instead of teaching analysts the same information individually, I thought it would be best to have a book for them to read. Then we could work on individual issues.
Ever learn anything from your students or come across something totally unexpected in your trainings?
Students usually use me as an on-the-spot consultant. They have been working on some problems for days, weeks, or months, and they ask me how to solve it in front of 100 people. Sometimes I can help, sometimes I can't. One of the last classes I taught before joining GE was for a military unit. They had me analyze live data in front of them, on the spot. That was interesting.
Where or who do you go to for inspiration in your writing?
I devote each book to someone in my family... Tao was for my wife Amy, Extrusion for my daughter Elise... now that I have a second daughter, Vivian, I will have to write a new book. I usually decide that I have enough original thoughts assembled, or enough material that no one else is addressing, and then I submit a proposal to a publisher. I have several on deck now (listed on my books page), but I will probably start a new project first that's not listed yet.
When I write I usually do it in long stretches that last from after work until 1 or 2 in the morning. I can do that once or twice a week, which is good for a chapter. I spend the rest of the week fixing what I wrote. After 12-20 weeks I have a draft manuscript, and copyediting begins. That can take 4-6 weeks. Production takes 4 months. So writing a whole book, if you add in about 2 months of preparation, can take about one year. That's very aggressive. I think if you take too long events pass by... unless you're writing THE ONE TRUE BOOK OF SECURITY FOR ALL TIME. :-)
You mentioned you were recently hired by General Electric to be the new director of Incident Response. What new challenges are you facing that you might not have been exposed to in your private consulting?
The biggest challenge is the scope of the company. The closest organization in size was the Air Force. Honestly, the challenge is the reason I was interested in the job in the first place.
How do you plan on tackling those challenges?
An obstacle I've encountered that needs to be overcome is the desire to avoid action because the company is too big. It's ok to take a sampling approach. With no data, you have nothing to investigate. With some data, you can make discoveries. Those bring investigative leads which direct additional data gathering. That is my plan for this challenge.
Over the next 2-3 years, what are the biggest challenges you think the security industry is going to face?
I see several challenges. If it hasn't happened already, people are going to wonder why they spent several million dollars deploying a SIM/SEM/SIEM, and they are "still being hacked." I am exceptionally worried about clients being exploited via Web browsers and subsequently controlled via encrypted Web channels. I see more of our investigation and security tools being directly targeted. I think people are going to spend millions on NAC and also ask why they are "still being hacked." I'm hearing that organizations with 80% NAC coverage are seeing intruders target the remaining 20%. Which, interestingly enough, are the most vulnerable already -- all those embedded Web servers on printers, routers, etc... along with new consumer equipment. There's no shortage of work.
During those years, what do you think we'll have to do right?
My constant plea is for more visibility. If you cannot see what is happening, you cannot make effective security decisions. To the extent you do anything right without visibility, you're just lucky. It's like playing goal with a blindfold. Visibility should be a design and deployment consideration, just like security is becoming.
More assets are being introduced to the enterprise. How do we deal with the risks introduced by threats who learn how to exploit vulnerable assets?
I return to visibility. The world changes too fast for anyone to understand it. If you can't understand it, the next best bet is to be able to watch it. If you can't watch it, how are you supposed to defend it -- or even know that an asset exists? I am not advocating abandoning your resistance mechanisms. (Notice I do not say "prevention." Prevention implies 100% effectiveness. Resistance implies that prevention eventually fails.) You should resist wherever possible but detect resistance failures. Sometimes that is only possible by keeping track of as much as you can, and then querying that data after-the-fact. Then you feed those lessons into your resistance mechanisms and repeat.
And finally, are we really still secure after all these years?
Nope! :-) Security is "the process of maintaining an acceptable level of perceived risk." It never ends.
Thank you Richard!
Thanks for the chance to share a few thoughts!blog comments powered by Disqus