Showing posts with label Security. Show all posts
Showing posts with label Security. Show all posts

Wednesday, September 7, 2011

The keys to the Kingdom


Privacy, or access restriction, is all about trust. Trust in the one guarding the access mechanism. The doors to heaven open but to the worthy, care of Saint Peter. A bank vault usually does not open until at least two different people get involved, each with their own key or code. The door to your home opens to yourself and perhaps one or two others you've given a key.

It's the same in the digital world. Computers and websites are accessed over connections, and these connections are vulnerable to trust-based attacks. Someone can easily get between you and your front door, so to speak. So called 'man in the middle' attacks involve a game of digital charades where a bad guy, Charlie, can spy on the exchange between Bob and Alice. Charlie simply pretends to be Bob when speaking to Alice and pretends to be Alice when speaking to Bob. Because it's all ones and zeroes, this works really well.

Now the common countermeasure is to encrypt the channel between Alice and Bob in such a way that a) Charlie has a tough time listening in and b) any attempt at impersonation is detected. This works by having a digital certificate, and some fancy non-invertible math. This approach is called the PKI system.

Basically, if I talk to you using the public key part of your digital certificate, only you understand what I'm saying. I would not get it myself if I heard it, post-encryption, but you do, because you can decode it with the private key part of your certificate. You can talk to me the same way, just look up my public key and encrypt your message with it. Only I will be able to reconstruct your original message. You're locking your message, and I have the only key.

Locks don't present much of a challenge to a locksmith. Digital locks like the PKI system I've just described have the same weakness, and this is where the trust comes in. Certificates are created by certificate authorities, and if you want to look up a public key, or verify that a sender is really the guy you think he is, you can check in with the authority. While surfing the web, your browser automatically takes care of this. The little lock symbol in your browser means that the connection is to the right entity and is secure.

This puts a lot of trust in the certificate authority. A certificate is only as safe as the authority is trusted. An evil or stupid CA could pass out your private key like candy, or lie to you about who you're connecting to, i.e. telling Alice she's listening to Bob when it's really our old pal Charlie, up to his tricks again.

The trustworthiness of authorities is the underlying issue of the recent discussion about certificates. Dutch CA Diginotar had the long arm of the Iranian secret service up where the sun doesn't shine, and was unaware of it. When they found out, they kept it mum while they were figuring out what to do. Bad idea. The only antidote to a compromised certificate or CA is to blacklist it immediately and install new, clean certificates. Anyone using the old ones is likely to have a Persian Charlie sitting in on his communications, and many did.

Diginotar is not just some two-bit CA. These guys have the keys to the Kingdom. Literally. Diginotar is one of the CA's who creates the certificates for the Dutch government, for major websites and services, and regrettably for a lot of Iranian dissidents too. They proved unworthy of this trust. The ramifications are huge. During the hack, a large number of certificates were created to compromise a wide array of websites and services. Browsers had to update their list of trusted CA's, the Dutch government stopped doing business with them (albeit late), and Diginotar's reputation is tarnished forever.  There's no knowing how much supposedly secure information has leaked while Diginotar was silent about the hack.

One benefit is that the security awareness of at least two countries was increased quite a bit. Now let's hope the government hires a better company instead of regulating the field some more. Trust is the coin of the cybersecurity realm, and it should be spent sparingly and wisely.

-

The (devastating) FoxIT report on the hack can be found here: http://www.rijksoverheid.nl/bestanden/documenten-en-publicaties/rapporten/2011/09/05/fox-it-operation-black-tulip/rapport-fox-it-operation-black-tulip-v1-0.pdf 


N.B. At the moment the site is down, probably due to the huge demand. Mirror here: http://tweakimg.net/files/upload/Operation+Black+Tulip+v1.0.pdf



Sunday, May 29, 2011

The deep end

DPI, or deep packet inspection, is a hot topic right now. Here in the Netherlands there's a raging debate about everything from net neutrality to privacy issues.

It started with Dutch mobile service providers publicly admitting that they are, and have been for a while, doing deep packet inspection on their customers' traffic1.

DPI is a technology for looking inside internet protocol packets to find out what kind of data is actually being transmitted. DPI can tell the difference between an e-mail and a tweet and a WhatsApp message, for example.

The mobile providers do this to find out if us sneaky mobile internet users are using services like voice-over-IP and online texting instead of paid-for regular phone calls and text messages.

As I've written before, I'm fairly square when it comes to property rights and fairly pessimistic where it concerns privacy. Deep packet inspection, and post-paying for bandwidth used on skype and WhatsApp troubles me deeply. I buy bandwidth for market prices do use as I see fit, and that's how the provider used to sell it.

The service providers are scared of loosing revenue and would rather conservatively repress free use of the bandwidth they are selling than go with the times and re-focus their business on what the consumer is asking for: freely useable bandwidth. History, it seems, has not elucidated obsoleteness enough, but the providers are generous in giving it ample opportunity time and again.

Now politicians have smelled controversy and votes, and several have taken side with the users to guarantee free use of bandwidth. Count on Dutch politicians to side with the little guy :).

Concurrently, four hours drive south of The Hague the G8 summit has been, by and large, advocating the direct opposite: monitor, inspect, regulate, police internet users and bill the bleep out of those nasty little consumers2. If they need privacy they are up to no good. If they share content they are stealing. If they use cheaper service is is a dangerous abuse of market self-regulation and everything should be done to protect obsolete revenue models. Guess how this is to be achieved? Right, DPI again.

So the guardians of intellectual property and the guardians of obsolete revenue models are united in a common goal: to inspect, log, and give consequence to the content of each internet protocol package going in and out of your computer, laptop and mobile, ever.

Interesting.

According to Cisco3, last year an average 237 petabytes was transmitted by mobile users per month. Good luck monitoring all that. That's a lot of packets to inspect. Will consumers be asked to field the price tag of their own repression? Higher prices, less functionality. Great business model.

Meanwhile Dutch minister Verhagen is industriously campaigning against the (ab)use of deep packet inspection by Dutch providers, citing privacy concerns. In his scenario regulation would be used for, rather than against the consumer's interests. Although I agree with his motivation (more votes privacy), I dislike the idea that the providers should be slapped with yet another regulation. My wallet can vote much louder than my ballot.

I'd like to think a daring provider will do the same thing to mobile bandwidth that Apple did to digital music: put a fair price on extensive usage rights, and re-invent the market. I'm not holding my breath though. Who'd dare go down the deep end?


1: http://www.engadget.com/2011/05/12/dutch-telco-kpn-using-deep-packet-inspection-to-monitor-mobile-c/

2: http://www.bbc.co.uk/news/technology-13553943

3: http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns705/ns827/
white_paper_c11-520862.html