Low And Slow, Persistence, Loud And Proud, And The Fundamentals

If you spend long enough in any business, then you see pretty much everything. Twice. I feel that way about my experience in security. In the olden days (as my kids say), it was about disruption and putting notches on the bedpost by taking down high-profile websites. The attacks du jour were disrupting Yahoo or E*Trade during the bubble days. That was hacking success.

Defending against this early Internet threat was as much about network architecture as it was about security. Diverse routes and other network layer defenses, combined with a new generation of high-performance firewall appliances (that wouldn't melt under a SYN Flood attack) neutralized this specific attack. Ah, the good, old days when you could buy a box to address your issues and speeds/feeds mattered. Remember those?

Then we saw the pendulum swing the other way, toward financial attacks and fraud. These attackers didn't want to be discovered, so they'd lurk and collect data that they could monetize either directly (by looting a bank account) or indirectly (by selling it). They mastered the low-and-slow attack, so your SIEM would roll the events, they'd go about their business, patiently and effectively, and you'd be none the wiser. Until you got a call from your payment processor, that is. Gonzalez was king of that world, taking hundreds of millions of accounts in one fell swoop. We can thank these guys for the rise of PCI to set a low -- OK, very low -- bar for security controls.

We can debate all day about whether the rise of financial fraud and the associated compliance regimes were good or bad for security. There are good arguments on both sides of the coin; given the size of the cyberfraud economy, this is an attack vector that is here to stay. But most of these attackers prey on simplistic attacks, stealing money the old fashioned way, one sucker at a time. It's hard to defend consumers from social engineering attacks. So we just increase the loss reserves and move on.

But now we are entering a new phase with multiple, new attackers with diametrically opposed objectives. On one hand, many organizations dealing with sensitive data and intellectual property -- typically in the military industrial complex -- have to deal with persistent attackers, the so-called advanced persistent threat (APT). Yes, my partner Rich Mogull says APT=China, and for the most part, he's right. But I think it's wrong to assume that other superpowers are not fielding a similar cyberintelligence gathering capability. And using it.

We've written a lot about these types of attacks on the Securosis blog, including what you can and can't do about them. Suffice it to say, if these persistent attackers want to be, they are already in your environment. The real question is what you do. Do you boot them out (so then they'll just find another way in)? Do you bury your head in the sand? Or do you try to put them in a box and perhaps even feed them disinformation? Those kinds of decisions used to be the purview of the military strategist. Now it's the purview of the CISO. How the times have changed, eh?

And as if that wasn't complicated enough, now we have another breed of attacker -- a more sophisticated version of the script kiddies that took down the major Web brands back in the day. Josh Corman calls these new attackers "Chaotic Actors," which is a decent characterization. These folks might be pushing for anarchy, political change, or just lulz, but in all cases they are brazen and the goal is embarrassment. They are loud and proud and aren't going away. They justify this behavior, saying the "real bad guys" already have your stuff, so they aren't really doing anything that wrong.

In fact, these guys combine the fairly simplistic attacks of the financial attackers (yes, SQLi lives on and on and on) with the objective of either taking down a site or looting them and dropping the goodies in the parking lot. And they might be right about the information already out there, but either way you now have a new set of attackers to worry about, with differing goals.

So what? There have always been multiple types of attackers aiming to do different things, so how is this different? Basically it's not, although everyone wants to focus on how this is new and dangerous and cause a panic. Now is not the time to panic. It's the time to get back to fundamentals: Things like monitoring and maybe even supplementing that with network full packet capture, so you can do more sophisticated analysis and forensics when these folks come knocking. Think about many of the tactics described by Professor Gene Spafford relative to being a bit more active in our defense. Don't forget about consistent user training. How many times do we need to see a high-profile company pwned by a low-level admin clicking on a compromised spreadsheet to get it right? Clearly a bit more.

It's time to refocus on egress filtering. We've spoken about the data breach triangle for years. You need valuable data, an exploit, and egress for a breach to happen. Since we have little control over data and exploits, we need to focus on egress. So do that. We also need to remember to close the simple holes. SQL injection is behind most of these new attacks. Come on, man! Any Web application scanner will find those holes. Then you need to systematically work with the developers to fix the issues, or deploy a Web application firewall to stop the bleeding.

Regardless of whether we are dealing with low and slow attackers, or folks whose egos require at loud and proud chest thumping, we can dramatically make it harder for them to be successful if we'd get back to the fundamentals. Not that they won't be successful if they are determined, targeted and persistent enough. But at least we won't be making it so damn easy.

Mike Rothman is president of Securosis and author of "The Pragmatic CSO."