Research

I just posted an explanation of Leopard Security (that’s Mac OS X 10.5 for you non-Apple geeks) up on TidBITS. It’s based on my original blog post here, but expanded and simplified to appeal to a more general audience. I realize I took some liberties with the explanations of buffer overflows, ASLR, vulnerabilities, and exploits, but I had to tailor the content for a less-security-geek audience. Check it out, and feel free to flame me here. I do believe that if everything works as advertised this is a very significant release. There are still some big holes (Quicktime anyone?), but Apple seems to be taking security more seriously than in the past few versions. Share:

IBM and Vormetric announces a deal yesterday where… well, I’ll let them say it: LAS VEGAS, NV – (MARKET WIRE) – 10/18/2007 – Vormetric, Inc. today announced that it has partnered with IBM to deliver database encryption capabilities for DB2 on Windows, Linux and Unix. IBM will offer Vormetric’s highly acclaimed data security solution as part of its data server portfolio, addressing customer demand for increased protection of sensitive data. This new capability is delivered in IBM Database Encryption Expert, initially available for the new DB2 9.5 “Viper 2” data server. First of all, I need to say I’m a big fan of Vormetric. They were the first distributed encryption product on the market and watching what they’ve done has really helped me evolve my thinking on enterprise class encryption. That said, I have a huge nit to pick with them over database encryption. Mostly because they don’t do it, at least as most think about database encryption. Vormetric is a file encryption product. A good one, with some cool additional features like user and application level access and encryption controls. but they don’t do field-level database encryption. Remember, I think encrypting the database files, especially when used with Database Activity Monitoring, is an extremely effective security control. But it doesn’t replace field level encryption, not in the long run. The role of file level encryption for databases is media protection first, with a little separation of duties second. It protects that database on disk and in backups. It also limits who can access the raw database files, but offers no protection for authorized users and administrators in the database. The role of field (column) level encryption is to provide separation of duties within the database. You can protect sensitive fields from those who have database access, including protection against database administrators. Two kinds of encryption. Two different roles. Two different problems solved. This is where I get annoyed with Vormetric’s (and now IBM’s) marketing. It confuses customers and tries to position file level encryption for databases as superior, instead of admitting that it solves a different problem. They seem to refuse to admit that field-level encryption plays a valid role in protecting database data. I realize it’s the job of their marketing to best position their product, but it’s my job to cut through the marketing and give you practical advice. Here it is: Vormetric does file encryption, which is a good option for media protection. Field level encryption is better for enforcing separation of duties, but since it’s hard to implement on certain systems you may need to start with file-level (preferably used with DAM) to buy you the time to migrate to field level. If you don’t need separation of duties, you don’t need field-level encryption, and file encryption is fine. I don’t like marketing that could place customers at risk or is designed to confuse the market, even when I like the product being marketed. Share:

Martin is on the road starting up his new job as a PCI auditor for Trustwave so I made my best attempt to record the podcast. More than a few technical difficulties later, we finally completed recording. Sorry about the extra reverb, I’m still figuring out my setup and accidentally left it a little high. For the record, Audio Hijack Pro rocks and I regret trying to record without it. The show is shorter tonight to account for Martin’s travel. We spend a fair bit of time talking about Apple products due to the upcoming release of OS X 10.5 and happenings in the world of the iPhone. I also chastise Martin for being in Denver and thinking the Rockies are just the big pointy things in the distance. I lived in Boulder for 16 years, and although I’m in Phoenix now I still tend to root for the old home teams. Except the Nuggets. Now if you’ll excuse me, I need to go pre-order Leopard for my Mac… Show Notes: OS X Leopard release and security features The Apple Store Rich’s commentary on Leopard iPhone Metasploit package HD Moore: Cracking the iPhone part 2.1 Rich’s commentary on the iPhone exploit Apple opening iphone, still scared of evil hax0rs Russian Business Network Citrix flaws or bad configuration Blame bad Citrix admins for poor site security, experts say Citrix; Owning the legitimate backdoor Sorry, no music tonight Network Security Podcast, Episode 81, October 17, 2007 Share:

No, this isn’t science fiction. According to Wired’s Danger Room, an automatic defense system went out of control in South Africa during a live fire exercise. Nine soldiers lost their lives, and fourteen were injured. I’m not going to make any jokes about this one, since we’ve crossed from the theoretical to the real, with a tragic loss of life. There’s not much else to say. Share:

One of the reasons I spend so much time talking about DLP around here is that it’s one of the first markets I covered as an analyst and I’ve been able to watch it grow from the start. It also means that over 5-6 years of coverage the odds are pretty high I’ve made some mistakes. The Usual Disclaimer: There are a lot of good DLP products on the market and I work with some of the companies. This post isn’t an explicit endorsement, and i’ll likely be highlighting competitors in future posts as they come out with their own product updates. Just keeping you informed, and you need to run through a full selection process to pick the best tool for your circumstances. With the strong rumors about the acquisition of Vontu, and since it was my first big mistake in this space, it’s a good time to come clean. Way back when Vontu was first coming to market they stopped off to meet me for lunch at the Walnut Brewery in Boulder, Colorado. I think I had a turkey burger because it’s only available at lunch, and I really like it. They described their key differentiator- using real database data to detect leaks, what they call Exact Data Matching (EDM). I wasn’t impressed, and informed them that Vericept could do it all with regular expressions. I walked away thinking I’d never see them again. A combination of factors proved me wrong. For the next 2 years Vericept didn’t recognize the value of the DLP market, continued to focus on acceptable use enforcement, and got their clocks cleaned by Vontu. A combination of aggressive execution, some key client references, and tight focus on leak prevention put Vontu in the top spot in the market. For the record, Vericept later brought in some new management that turned the company around, putting them in second place in terms of revenue by last year. Nice thing about an early market, you can afford some mistakes. Most customers still don’t use EDM, but that’s not the point. I thought, at the time, that a general platform would be more successful, but it was the focused solution that clients were more interested in. Even if the Symantec deal doesn’t happen, that laser focus on the business problem has already paid off. The next example of poor judgement concerns Reconnex. Reconnex is unique in the DLP market in that they can collect all traffic, not just policy violations. I used to call this full forensics since it was essentially structured network forensics. Back when they released the first versions of the product this feature wasn’t an advantage for DLP. There was no reason to collect all that traffic; sure, it might be helpful in an investigation, but few DLP clients were interested. Management at the time (since changed) focused so much on that feature that they let the user interface and performance slack. With their new release, I may be changing my mind. They’ve now turned the capture capability from a forensics tool into a data mining and policy validation tool. Aside from still being useful in investigations, you can now generate a DLP policy and run it on old data. Instead of having to tune a policy in production as you go, you can tune it offline and play with changes without affecting production. They’ve also added data mining so you can use the tool to help identify sensitive data that’s not currently protected by a policy by looking at behavior/history. I haven’t talked to any references about this yet, but it looks promising. They’ve also revamped the user interface and it’s much more usable with better workflow. I know some of the other DLP vendors are working up their next releases and it will be interesting to see what pops. I’ve already heard some good things about the endpoint capabilities of one of them, although they haven’t briefed me. Share:

Honey? My Blackberry broke. What? I don’t know, it just stopped working. Yeah, I know it looks like it fell off the roof, but I don’t know how that could have happened. Okay, I’ll still probably wait for a 3G version since I really like my Blackberry Pearl, but this is an awesome move. I will, however, call bubkis on this next part: Apple “[is] excited about creating a vibrant third party developer community around the iPhone and enabling hundreds of new applications for our users,” but they are taking the time to do it properly “because we’re trying to do two diametrically opposed things at once – provide an advanced and open platform to developers while at the same time protect iPhone users from viruses, malware, privacy attacks, etc.” Wait, last time I checked the Mac was an open platform, relatively safe from “viruses, malware, privacy attacks, etc.”? And doesn’t the iPhone run on OS X? Last time I asked those questions the response was… a little chilly. Updated: Glenn over at TidBITS predicted this last week. Great scoop! Share:

As rmogull. Adam Engst got me started with this article. Seems more useful than I expected. I’ve added it to the contact links on the home page of the blog. Share:

Hoff (and some others) have been talking a lot about hope and the future. Chris has dedicated most of his recent posts to making us think differently about security. To drop our archaic models of the past and look towards solutions for the future. It’s a noble goal, one I support completely. Dr. Eugene Spafford, a seminal figure in information security, is also dedicating effort to the cause. I’m firmly in their camp and believe that while we don’t need an entirely new model for security, we definitely need to evolve. Information Security has been little more than basic network security and antivirus ever since Code Red and Melissa hit. But that’s not important right now. The essential questions are, “will we win?” And “do we make a difference?” These questions are non-trivial and endemic to the human condition. Anyone, in any occupation, who is invested in what they do will frequently use these questions to position themselves in the world. For some an occupation is merely a way to pass the hours and pay the bills; these automatons contribute to the status quo, but don’t help society evolve. For the rest of us our occupation is an essential component of our identity. We define ourselves by our occupation, and define our occupation as we want to define ourselves. I’ve worked in public safety my entire adult life, and spent most of my childhood, purposefully or not, preparing for my strange career. Over the years as I worked in different positions throughout public safety, from physical security, to emergency medicine, to information security, I was challenged by difficult questions of conscience. When I started in emergency medicine, I had to reconcile the thrill of the job with the fact that I achieved professional satisfaction only through the pain and suffering of others. As much as I wanted to try that new procedure, or be on that big call, I had to accept that for me to exercise my skills, someone needed to suffer injury or illness. I reconciled such a potentially twisted mentality by realizing that it wasn’t that I wanted someone else to suffer, but I wanted to do my job and do it well. People will get hurt, sick, and die with or without my involvement; I was a professional and wanted to do the job I was highly trained for. If something was going to happen, I wanted to be the one to be there. As my experience and confidence grew, I also began to believe that the better I was at my job, the less that victim (or the family) would suffer. Physical security was similar, but involved some slightly more complex mental gymnastics, which every cop and (I expect) soldier experiences. While as a medic you relieve pain and suffering, in physical security you often inflict it. We all loved the rush of breaking up a fight or catching a bad guy. There is an undeniable thrill in being authorized to use physical force on another human being- not a thrill of sadism, but the same emotions evoked by the sports we use to sublimate physical combat. In those cases my goals became to use as little force as possible and de-escalate situations verbally. Violence was not the objective; it was the last tool available to protect others. I’d like to call it altruism, but the truth is there are visceral thrills and deep satisfaction in managing the challenges of emergency medicine, rescue, and physical security. I learned to accept this motivation without guilt, since the goals of safety and security called for such commitment. When safety and security become excuses to do bad things, that’s when a very bad line is crossed. But back to security. In information security we may not be faced by the prospects of blood and guts, but those of us “in the industry” need to accept that we make our money off the pain of others. There’s nothing wrong with this so long as we don’t take advantage of our clients. I’m not just talking about vendors; we in internal security also provide a service to a client. My personal philosophy around this is that I won’t lie or try to frighten just to enhance my own income, but I’ll tell the truth and charge what I think is fair value for my services. I also still perform some volunteer work for those who need the help but can’t afford it. Security professionals earn our daily bread from fear and pain (sometimes very abstract pain, but pain nonetheless). There’s nothing wrong with that, but it does convey a responsibility not seen in other occupations. The big question I haven’t addressed, one that underlies pretty much any occupation, is, “Do I make a difference?” Psychologically I believe all humans fundamentally need to make a difference. It’s hard wired into our brains. If we’re not making a difference, we have only one of a few possible reactions. We can disengage from that activity and find fulfillment in other parts of our lives, or disengage from life completely. As sad as that sounds, we all know people who don’t see the meaning of their life and instead turn to a never-ending trail of distractions. We can also deceive ourselves and create illusions that we matter; I suspect many mountains of bureaucracy have been built on such falsehoods. We can also seek satisfaction elsewhere; actively finding a new job or career. We can also do the absolute best job possible, fight the good fight, and try to rise above any limiting circumstances. As a paramedic I may have been the one who saved a few lives and reduced a little suffering, but the reality is that if I hadn’t been there, someone else would have been. In mountain rescue we operate as a team and it’s a group of 40 or so people, not some lone hero, that makes the save. But although I personally wasn’t essential, and the rescue would have happened

This is such a straightforward problem to solve it’s annoying that it still makes the headlines. Laptop and tape encryption are the low hanging fruit of data security. Not that they are click-box easy, but it’s pretty straightforward for most organizations to protect this stuff. Home Depot lost a “password protected” laptop when it was stolen from a car, and 10,000 employee records with it. Iron Mountain lost a case of backup tapes with a decade’s worth of Social Security Numbers from college applicants in Louisiana. Their proactive strategy to protect their customers? “We certainly don’t want to create any panic. But people should be aware and take the necessary steps,” Amrhein told the AP. “This is backup data off of a mainframe that contains sensitive personal information.” Darn, it’s my fault for applying to college and not being aware. Silly me. I do take umbrage at some of the misguided advice at the end of the article: “If you buy encryption you need to work with the company’s legal department and top executives on a process where you can prove data on a stolen device can’t be tampered with,” he said. “A cradle-to-gave transaction record on the server is one way to provide an inventory on the current state of all your drives. Another, more difficult approach is to write everything down.” He said it helps if a company can show it is using a reputable vendor to put a barrier around stored data, and mentioned Seagate Technology as an example. The Scotts Valley, Calif.-based hard drive maker said this week it will roll out enterprise-class drives with full disk encryption in 2008 and will push to make hard-drive encryption standards a reality to reduce complexities that could hinder adoption. Like a cradle to grave transaction record and an inventory of all you hard drives is realistic. Also, while encrypted drives will play a role in data security they are far from a panacea! First of all, the software solutions today, especially for whole drive, are effective without requiring you to install new drives. Second, the encryption on those drives is managed by software, so now you’ll have to buy both the encrypted drive and the software to manage it. More often than not, non-laptop encrypted drives are totally unnecessary and don’t improve security. I like how Seagate designed their drives, but it’s not like they’re the right choice in all cases, nor will they put us (or software encryption) out of business. Remember the Three Laws people. Use your encryption well. Share:

Apple has finally released the full list of updates in the next version of the Mac operating system, including a section detailing all the security updates. A couple of features look pretty interesting. The biggest is the inclusion of “Library Randomization”, or what we call layout randomization (ASLR) in Vista. System functions are randomized in memory to make exploitation more difficult. I don’t have a Leopard seed to check it out, and I suspect some of the researchers out there will dig in and let us know how good (or bad) the implementation is. Mac OS X already supports Data Execution Prevention, one of the other key Windows XP, Server, and Vista anti-exploitation technologies. Another good feature is tagging of downloaded applications. Any downloaded executable is tagged by the OS and requires the user to approve it on first launch (it doesn’t mention if it’s a password prompt or just clicking an OK box). It appears to list the app name, what tool downloaded it, and (if possible) the URL it came from. Regular users probably won’t pay attention, but this will be nice for those of us who do. Apple also (finally) improved the Mac OS X firewall to include some level of application control. The description makes it look like it only controls inbound connections, which would be too bad. I think the user interface for this one will be pretty important, and maybe outbound control is hidden in the capabilities somewhere. Anyone up to date on ipfw that can let us know if Apple is sticking with that? There’s a new sandboxing feature for some default applications, including Bonjour, Spotlight, and Quick Look. I highly suspect this is a way of limiting the potential exploitation via file and network fuzzing, considering the applications they picked. Most of the rest of the updates are fairly straightforward and good to see. Application signing, 256 bit AES for file encryption, better VPN support, SMB packet signing for Windows compatibility, multiple user certificates, and some updates to access control lists for file sharing (I think, although they don’t say, driven by Windows compatibility issues). There’s increased smart card support designed to meet the needs of the feds, but I might give it a shot (for fun) if the readers are added to default Macs (unlikely). And let’s not forget the biggest security feature in Leopard that didn’t make the list- Time Machine. Getting users to do differential backups will do more to assure the availability of their data than any other security feature. I’m really looking forward to seeing how this all holds up once the security researchers get their hands on it. On paper it looks great, maybe even getting Mac OS X up to the level of Vista (for security- usability on Vista still sucks). But I don’t believe anything until people smarter than me start banging on it and seeing where the cracks are. Share: