DNS inventor blames wrangling for insecure interweb
Mockapetris slams 10 years of 'political and technical dithering'
The inventor of the domain name system has blamed technical and political wrangling for delays in improving internet infrastructure security.
Dr Paul Mockapetris, chairman of DNS firm Nominum, explained that the use of digital signatures would help fix fundamental vulnerabilities in the internet's domain name system, highlighted by a much-publicised flaw discovered by security researcher Dan Kaminsky this summer. The flaw created a means to redirect surfers to potentially malicious sites by poisoning DNS look-up tables.
The fix for the flaw, yet to be applied by many, involves source-port randomisation, but this is only a partial fix for a much bigger problem, according to Mockapetris. He reckons that the DNS systems on high speed networks might still be cracked in five hours by an attacker with sufficient knowledge and resources.
"An attack might be possible in five hours with the patch. That's much better than the minutes an attack might have taken before but systems are still not really protected. IT's bought time without solving the underlying problem," Mockapetris told El Reg.
DNSSec (Domain Name System Security Extension), which uses digital signatures to guard against forged requests, offers a means of making internet naming systems more secure. But even 15 years after the standard was developed its adoption remains low.
Mockapetris blames problems in making the technology easy to deploy, delays in developing DNSSec-aware apps, and political wrangles in holding back adoption of the technology. Arguments about whether or not to give VeriSign the role of a trusted third party signing root keys have acted as a roadblock but Mockapetris reckons difficulties in making the technology easy to apply are the greatest obstacle to its deployment.
"There were five years of good work in there to roll out the technology but on top of that we've had 10 years of political and technical dithering," Mockapetris said.
Only a massive blockbuster attack or applications that require DNSSec are likely to spur adoption of the technology, which has never really got out of first gear. One such application could be using using DNS data to distribute reputation on domains taken from feeds such as Spamhaus and blocklists maintained by Trend Micro.
URL filtering and content filtering technologies do a similar job but Mockapetris explained that using the DNS infrastructure would offer faster response alongside greater agility, for example, in tracking botnet systems using Fast Flux approaches to hide phishing and malware sites behind compromised proxy hosts.
Nominum has added MDR (malicious domain redirection) technology to its proprietary DNS technology as a way for ISPs to block access to child pornography sites, based on IP address.
Mockapetris helped invent the DNS system in the 1980s. The original intention was to get systems up and running and add security features later but the process has proved far more protracted than he ever imagined. Mockapetris now reckons DNSSec might eventually be applied in 2015 but given he said five years ago that the technology would be "ubiquitous" by 2008 we ought to treat such predictions with caution.
Whenever a more secure alternative is applied Mockapetris will be recognised as doing as much as anyone to promote its adoption. As well as being one of the parents of DNS, alongside the late Jon Postel, Mockapetris might then also be recognised as one of the midwives of DNSSec. ®
I guess what us end-users with our own DNS servers need is a patch that allows trusted stuff inside the firewall to carry on as before, querying the local DNS server that will then use DNSSec out on the Wild Wild Web^WInternet.
So, as asked by others, is there a secure public service available we can use. More to the point, *can* there be a secure public service available until all authoritative servers support it? Otherwise how does my DNSSec server know that it's talking to the real server for the site I'm trying to access?
I'm ready for both just as soon as the rest of the 'Net is - just show me the kit!
Coat, as I can't see either happening any time soon...
Not totally sure DNSSEC was ready enough to be deployed technically 15 years ago by many people and some improvements have been made since then but not enough to make deployment at all easy. Another factor holding back deployment is the Esperanto problem. Great language, but too few people speak it to make it worth learning. The application that will probably need to go to DNSSEC first is probably money - banks are likely to adopt DNSSEC themselves and then force e-commerce sites to use it next because the losses from phishers routinely compromising ordinary DNS become too great.
Once enough ISPs support DNSSEC for their customers, your bank won't let you make on-line payments or do online banking from ISP connections that don't.