Gadfly zeroes in on Oracle bugs
update The Oracle community, says David Litchfield, shouldn't be so smug when it comes to database security.
Editors' note: After this story was published, the reporter asked David Litchfield follow-up questions about ties to Microsoft and Oracle upon learning that Microsoft was a customer of NGS Software.
update ARLINGTON, Va.--Don't even try to tell David Litchfield that Oracle is unbreakable.
Litchfield, a noted bug hunter, has made it his mission to tell the world that database software is insecure--Oracle's database software in particular. Litchfield has been vocal in his criticism of Oracle, even calling for the resignation of Oracle Chief Security Officer Mary Ann Davidson.
For too long, Oracle and its customers have
Litchfield, co-founder of Next Generation Security Software in the U.K., is on a crusade. In January he published The Oracle Hacker's Handbook. The book, according to its cover, offers readers a complete arsenal to assess and defend Oracle systems.
While dissing Oracle, Litchfield is cheerleading for Microsoft. He has publicly stated that SQL Server 2005, the latest version of Microsoft's database software, is secure. This must hurt at Oracle, a Microsoft arch rival, which has already seen a significant piece of the database market go to the Redmond, Wash.-based software giant.
When not hunting for bugs, Litchfield likes to go out with his two greyhounds, and he helps charities find homes for other canines. In fact, he is so passionate about his dogs that he dedicated The Oracle Hacker's Handbook to his wife and two girls, the girls being his greyhounds.
At last week's Black Hat DC event, Litchfield discussed a new attack technique that increases the severity of certain vulnerabilities in Oracle's database software. He sat down with CNET News.com at the event to explain why such disclosures are necessary.
Q: Why are you into database security? There's so much other software out there.
Litchfield: Because that's where the crown jewels are for any organization. Every organization on this planet has a database and that's where the lifeblood of that organization exists. Where better to secure it than at the source. We can secure it at the perimeter, but with vulnerabilities like SQL injection, that security is completely undone.
Despite having a firewall, despite having the Web server locked down, an SQL injection flaw in your Web app takes us all the way through to the back end of the database server. If that database is not using the principle of least privilege or is not fully patched, then we can gain full access to the database server and suck out all your data. The database has to be secured. The problem is that nobody has ever really dealt with the back end until recently. It has always been about securing the perimeter.
Lately you have especially been looking closely at Oracle's databases. Is there a specific reason that you're looking at Oracle more than Microsoft or IBM?
Litchfield: Yes. SQL Server 2005 is secure. (Microsoft has) solved the problem. Oracle is in the process of solving that problem. IBM, I have looked at DB2 and Informix and sent them a bunch of bugs, probably about 50, ranging from buffer overflows to privilege escalation issues. But IBM's security response was mature. In the most recent past, the Oracle security response was not so mature. They have been combative, as opposed to: "This guy is just trying to make our products more secure." But it is getting better. Oracle is beginning to understand that we're fighting on the same side, just from different perspectives.
When a vendor like Oracle becomes more combative, you become more combative as well?
Litchfield: I will. It is unfortunate that it happens that way, but if you have to defend yourself, then you should defend yourself. I would rather be working, like I do with Microsoft and IBM, with their security response team. We've got good relationships with Microsoft and IBM. What better way to get things done than have a good relationship, as opposed to sniping at each other from the gutter.
My relationship has gotten slightly better with Oracle, and they understand that it's not so much a battle of wills. I'm trying to make them aware of these problems in their database because it affects me indirectly. If someone breaks into that database server and steals my information, then I'm paying for it, not Oracle.
Some might think that it's some sort of an extortion game that's being played.
Litchfield: I've never asked Oracle for money. If people think that, they are ill-informed.
And Microsoft doesn't pay you to say SQL Server 2005 is secure.
Litchfield: I'm not being paid by Microsoft to say they're secure, and if anyone is going find a bug in SQL Server 2005, it better be me. It would undermine my ability to be able to say in the future that a product is secure if bugs are found by anyone else. So, if there are bugs in SQL Server 2005, I hope I'm the one who finds them, and I'm looking.
Is Microsoft, or was it ever, a customer of NGS Software?
Litchfield: Whilst NGS does do a lot of work for Microsoft, we're not paid to say they're secure--we're paid to help secure their products. It's important for both Microsoft and NGS that we are, and are seen to be, independent. Otherwise it would call into question the legitimacy of the work we do and Microsoft's efforts to secure their products. This is why NGS still continues to generate (security) advisories on Microsoft products.
What's the business of NGS Software?
Litchfield: There are three sides to the business. We sell tools to help assess your state of vulnerability and whether you're compliant with Sarbanes-Oxley, etc. We consult to a number of organizations, and we also do vulnerability research and sell that research.
What types of organizations are your typical research customers?
Litchfield: Government organizations, those who are responsible for critical national infrastructure and the protection thereof. We try to give them advance warning of security problems. We can tell them that there's a flaw in a particular product, along with a risk mitigation strategy. Even in the absence of a vendor-supplied patch, these systems will be protected.
NGS has been growing over the past years. Where is the demand?
Litchfield: It's mostly in consultancy, which is a bit of a shame because I set out to build a software company and we're more of a consultancy. That's one of my personal failures though and I've not given up. We will be a software company at some stage.
What does a typical consultant do?
Litchfield: He might do penetration tests, code reviews or threat modeling. It is not installing firewalls; our consultants do the high-end stuff.
What is it that drives you to get up and do your job every day?
Litchfield: Well, I'm good at it, and if you're good at something, you've got more drive to do. If I was a good painter, I would paint more, you know. But since I am crap, I don't bother doing that. I enjoy the work.
Particularly the bug hunting part of it?
Litchfield: Yes, it's just a question of analysis. If I were trying to subvert the system, how would I do it? The other reason is that it has an effect on everyone's lives. Now, it's not like I'm curing cancer, but I know that one database server tomorrow is going to be more secure because of something I did, and that means that, for example, more credit card numbers are safe that day.
If people at Oracle say that you actually hurt security because of the disclosure of vulnerabilities, what do you say?
Litchfield: They have a case. In certain cases it does raise the risk level, OK, and that's one of the major problems with this kind of work. However, in raising the risk level, people are more inclined to protect their systems.
Now, as an example of this, I just put out that new attack method that allows an attacker without any special privileges to exploit flaws that were thought to be only exploitable by people with higher privileges. Now we know that that's not true, people have no reason to say they are not going to patch.
Someone has, within day zero of me posting my new method, modified their exploits and posted them publicly to use my new methods. Those exploits now can be run by anyone. So, yes it has increased the risk.
Back in August 2002, I presented some code that was then taken to form the basis of SQL Slammer. There was that initial raising of risk, but after that short-term pain, there are now more patched SQL Servers out there than there ever were before. The short-term risk has been raised for the long-term benefit, that's the way I look at it.
If people like you weren't around, some might say we wouldn't know of any security risks and nobody would be exploiting them either. You don't think that's true?
Litchfield: I don't think that's true. There are always going to be bad guys out there. If there aren't good guys working with the vendors to close these holes, then we'll be walking around with our head in the clouds thinking we're all secure when we're not. Security through ignorance really doesn't work because one person's ignorance is someone else's revenue stream.
What makes you pull your hair out?
Litchfield: When people say things like I'm increasing risk or doing it for selfish reasons. I'm not like that. But I can't always be the popular guy. I just wish there were fewer detractors.
You published The Oracle Hacker's Handbook recently. What do you hope to achieve with the book?
Litchfield: The Oracle security world is smug basically, and I'm trying to take that security blanket away from them. There are too many people out there who think that the Oracle product is secure and that they don't need to be doing anything. That's irresponsible, as far as I'm concerned.
What would you like people to know you for?
Litchfield: I would like to be the person who helped convince people that database security is important to look at. I would like to think that it's through my work, and obviously that of some of my peers in the industry, that we've helped shape the way security is dealt with at places like Oracle and Microsoft.
