- Related Stories
-
New denial-of-service threat emerges
March 16, 2006 -
Skype could provide botnet controls
January 25, 2006 -
Good security news to be in short supply in 2006
January 20, 2006 -
Blackmailers try to black out Million Dollar Homepage
January 18, 2006 -
Bots slim down to get tough
November 16, 2005 -
Kevin Mitnick on hacking's evolution
November 4, 2005 -
Old software weakening Net's backbone, survey says
October 25, 2005 -
The sorry state of the domain name game
October 4, 2005 -
Weak links in the Net's armor
August 3, 2005 -
Hacking for dollars
July 6, 2005 -
VeriSign to put more backbone into the Net
May 19, 2005 -
Report: Crooks behind more Net attacks
November 16, 2004 -
Blackout hits major Web sites
June 15, 2004 -
Jerry Garcia's guitars up for auction
May 6, 2002
(continued from previous page)
It is generally possible to stop the more-common bot-delivered attack by blocking traffic from the attacking machines, which are identifiable. But blocking queries from DNS servers brings problems in its wake. A DNS server has a valid role to play in the workings of the Internet. Blocking traffic to a DNS server could also mean blocking legitimate users from sending e-mail or visiting a Web site.
Lowdown on Net threat
Key things to know about distributed denial-of-service attacks sent using recursive DNS queries.
Why can DNS be abused for a DDOS attack?
As many as 75 percent of all DNS servers are open to requests from anyone on the Net, a feature called "recursive queries."
How does such an attack work?
A hacker uses a botnet to send a large number of DNS queries to open DNS servers. These queries are "spoofed" to look like they come from the target of the attack. The DNS servers that receive the query will reply to that network address.
How does this amplify the attack?
A small DNS request sent out by a bot can result in a much larger response from the DNS server. The amplification factor can reach a factor higher than 70.
Why does abusing DNS protect the attacker?
The malicious traffic seen by the victim now no longer comes from the hacker's botnet, but from DNS servers. This could help hide the network of zombie PCs.
Why can't the malicious traffic be blocked?
DNS servers are play a vital role on the Internet. If a company blocks its DNS server, legitimate users may no longer be able to send e-mail or surf to its Web site.
What can be done to limit these attacks?
DNS administrators should disable the recursive functionality or limit it to their own, trusted users. Also, targets of such attacks could protect themselves using technologies designed to ward them off.
Source: "DNS Amplification Attacks" by Randal Vaughn and Gadi Evron, CNET News.com research
"That's why this is a nasty attack," said Rob Fleischman, the chief technology officer at Simplicita, a Denver-based security start-up. "The DNS system is an area that is going to be under more attack. It is going to have closer scrutiny and more security."
At the heart of the problem are so-called recursive name servers, which are DNS servers that allow queries from anyone on the Net. There are about 7.5 million DNS servers, and estimates on how many are left wide open to queries range from 600,000 to 5.6 million, according to Vaughn and Evron's report.
"People who are running these open servers need to clean up their act. They are--witting or unwitting, lazy or just don't care--participants in these attacks," Mockapetris said. "They are the Typhoid Marys of the Internet."
To protect their systems, organizations with DNS servers can disable the recursive feature that lets anyone look up addresses. Alternatively, they can manage the server settings so that the recursive feature is available only to insiders. Internet service providers, as well as businesses and individuals, are among those who run DNS servers.
Targets of DDOS attacks could protect themselves using technologies to ward of DDOS attacks, which are sold by vendors including Prolexic Technologies.
In the early days of the Internet, recursive DNS servers served mobile users and cached people's requests for Web site addresses, making the Net scale much better, Mockapetris said. An example of the latter was the day Jerry Garcia died in 1995, he said.
"Everybody was going off to find every Grateful Dead Web site everywhere in the world," he said. "The first person to do that would cache it in the DNS server of their access provider, so the next person would not have to go out to Katmandu to look it up."
But fast forward 10 years, and recursive servers should be something of the past, Mockapetris said. "Now people are looking for ways to attack the network, and the open recursive servers can be used as unwitting cat's paws in a denial-of-service attack," he said. "Once upon a time, everybody just trusted everybody, and you would say, 'Fine, use my server.' Now you have to be more careful about that."
Kaminsky agreed. "If you are a DNS administrator, you shouldn't be providing recursive services to the Internet anymore. It is unfortunately no longer a responsible thing to do," he said.
Increasingly, DNS is going to be used in attacks, experts said, and their administrators can no longer afford to be lazy.
"There are multiple of these kinds of storms that are rising, and service providers and enterprises need to figure out how to make sure that their sea walls, dams and dikes and levees are high enough to withstand them," Mockapetris said.
See more CNET content tagged:
DNS server, Nominum Inc., DNS, distributed denial of service, VeriSign Inc.






how do they actually detect pharming? checking the domain name certainly doesn't do it, and not their 52 rules of checking (hey why not make it 521 rules, it sounds better)
I wouldn't waste $ on this crap if I were you.
http://www.google.co.uk/search?hl=en&q=Akamia+attack&meta=
http://www.google.co.uk/search?hl=en&q=Akamai+attack+n3td3v&spell=1
2. And theres still the Yahoo Slurp disclosure that Gadi and the others can't work out.
I guess that'll come up two years later too...
When folks like Gadi on FD can't work something out on FD, they call you a troll or tell you to goto school.
Its funny.
irresponsible, I think it highlights that we
need to move beyond IPv4 to something that
doesn't just assume a connection is legit. If a
connection is made apparently from ___, then the
ACK of receipt used in file transfers (i.e.,
packet received) could serve as a "you did send
this, right?". IPv6 has some improvements over
IPv4 as well.
So think.
See http://blogs.jhsoft.com/jhsoft/PermaLink,guid,f43ae4a8-b3cb-43ba-b9c0-261f4a4b509c.aspx
Anyone know of anything else?
Reverse lookups performed by hardware allow the latest Cisco devices (those employed with the Supervisor 32 or Supervisor 720 module) can determine whether the sender is valid or spoofed and it will drop spoofed DNS requests.
That said... this article should have been written with the title... "For those not willing to invest in proper protection... DNS continues to plage them!!!
The problem has been known for several years adn a valid solution exists... but it's only for those whom invest properly in the correct security equipment!!!
Walt
- Problem has been solved...
- by wbenton March 26, 2006 6:54 AM PST
- For those whom have invested in the proper hardware and software from Cisco... this problem has been forwarned for several years now and Cisco has a resolution.
- Like this Reply to this comment
-
-
- Not really...
- by rdeutch March 30, 2006 2:09 AM PST
- If the Cisco device does a reverse DNS lookup on the spoofed IP address, it will still get a correct result. So how does this solve anything?
- Like this
-
(16 Comments)Reverse lookups performed by hardware allow the latest Cisco devices (those employed with the Supervisor 32 or Supervisor 720 module) can determine whether the sender is valid or spoofed and it will drop spoofed DNS requests.
That said... this article should have been written with the title... "For those not willing to invest in proper protection... DNS continues to plage them!!!
The problem has been known for several years adn a valid solution exists... but it's only for those whom invest properly in the correct security equipment!!!
Walt
If an IP packet with a spoofed origin IP address reaches its target (or the firewall in front of it), then there is no way to tell if the packet really came from the claimed IP address or not.