Daft users and insecure web apps dominate threat index
Software vulns are so last year
Cyber criminals and spies have shifted their focus of attack in response to improved security defences.
Facing improvements in system and network security, crackers have two new prime targets that allow them to evade firewalls, anti-virus, and even intrusion prevention tools: users who are easily misled and custom-built applications, according to the latest annual threat landscape report by the SANS Institute.
The latest edition of the SANS Institute's Top 20 Internet Security Risks list, published Tuesday, highlights a shift away from traditional avenues of attack against flaws in commonly used software packages towards more customised and targeted assaults. Although the Top 20 focuses on emerging attack patterns, old-school vulnerabilities remain a problem.
Browser security bugs and the like are still being targeted by automated attack programs that scan the web for vulnerable systems. A new system can expect to survive only five minutes on the net before being attacked, according to experts at the SANS Institute's Internet Storm Centre.
Qualys, which markets tools that scan for vulnerabilities, reports a "huge jump" in the vulnerabilities in Microsoft Office products, up 300 per cent over the last 12 months. Excel vulnerabilities were the main factor in this growth, according to Amol Sawarte, manager of the vulnerability labs at Qualys.
Patching and standard defences (such as firewalls and intrusion detection) tools go a long way towards fighting off attacks against software vulnerabilities. Defending against user stupidity or attacks against customised applications is a much harder task.
"For most large and sensitive organisations, the newest risks are the ones causing the most trouble," said Alan Paller, director of research at SANS. "The new risks are much harder to defend; they take a level of commitment to continuous monitoring and uncompromising adherence to policy with real penalties, that only the largest banks and most sensitive military organisations have, so far, been willing to implement."
Web application security is a particularly thorny issue because many developers know little about security. Once breached, web applications provide a handy avenue into to back-end databases that hold sensitive information.
"Until colleges that teach programmers and companies that employ programmers ensure that developers learn secure coding, and until those employers ensure that they work in an effective secure development life cycle, we will continue to see major vulnerabilities in nearly half of all web applications," Paller said.
Forty-three security experts from government, industry, and academia in a half dozen countries cooperated to produce the Top 20 threat list of the worst security risks. Suppliers agree that securing web applications is among the toughest challenges facing the industry.
"Although half the total vulnerabilities reported in 2007 are in web applications, it’s only the tip-of-the-iceberg," said Rohit Dhamankar, senior manager of security research for TippingPoint. "These data exclude vulnerabilities in custom developed web applications. Compromised websites provide avenues for massive client-side compromises via web browser, office documents and media player exploits. This vicious circle of compromise is proving to be harder to break each day".
As in past years, Qualys has released a tool (registration required) that allows users to test computers for the elements on the Top 20 that lend themselves to remote testing.
Best practices for preventing top 20 risks:
- Configure systems, from the outset, with the most secure configuration available from the vendor and use automation to keep users from installing/uninstalling software.
- Use automation to make sure systems maintain their secure configuration and remain fully patched with the latest software versions and anti-virus updates.
- Proxy critical client level services.
- Protect sensitive data through encryption, access controls, and automated data leakage protection.
- Use automated innoculation for awareness, and establish penalties for those who do not follow acceptable use policy.
- Perform proper DMZ segmentation using firewalls.
- Push to minimise the security flaws in web applications by testing programmers' security knowledge and testing the software for flaws.
In other words, as SANS summarises, trust but verify through automation and testing. ®
Careful with that, ensure flame retardent coat is donned!
Well, yes - comparitively speaking web based applications will always be the weak point in the security chain - they have to be known, public locations.
The trick, I reckon, is to work out what your web-app needs to know. Does it need access to the back office? Really (the answer here should be "no")? How much user data does it need and how much should be encrypted? How are you maintaining sessions and preventing hijacking? Does it matter if the session is hijacked ("if you are not Chris Cheale - click here" Amazon style)? This rather depends on what that session allows access to; is it a "public" or "private" session?
Another thing is how well your app cleans up after itself; just how good is the garbage collection? Don't leave dead session (or other) data lying around - it's just asking for trouble. Oh, and trust no-one - sanitise all data going between your application and display layers - type fix it where you can.
What worries me a little is that because I'm an entirely self-taught LAMP-type developer, I'm sure there are things I'm missing - so I never stop trying to learn.
What worries me a _lot_ is the amount of code for web-apps I've seen that is utter shite; that I'd be ashamed to put my name to, let alone charge anyone for.