Krebs DDoS aftermath: industry in shock at size, depth and complexity of attack

It's not the size that's most worrying Akamai and the anti-DDoS industry so much as its depth, form and complexity

Last week's historic 620Gbps+ DDoS attack on the website of security blogger Brian Krebs was no ordinary attack, according to the firm that found itself on the receiving, end, Akamai.

The peak level, which hit Akamai for most of Tuesday and Thursday, was probably somewhere between 600Gbps and 700Gbps, more than enough to cause major problems both to Akamai and fellow peers that start dropping general Internet traffic at that level.

That is big, massive in fact, but it's not the size that's most worrying Akamai and the anti-DDoS industry so much as its depth, form and complexity.

"This attack didn't stop, it came in wave after wave, hundreds of millions of packets per second," says Josh Shaul, Akamai's vice president of product management, when Techworld spoke to him.

"This was different from anything we've ever seen before in our history of DDoS attacks. They hit our systems pretty hard."

Clearly still a bit stunned, Shaul describes the Krebs DDoS as unprecedented. Unlike previous large DDoS attacks such as the infamous one carried out on cyber-campaign group Spamhaus in 2013, this one did not use fancy amplification or reflection to muster its traffic. It was straight packet assault from the old school.

"This was an enormous botnet. This was a straight attack. It was astounding how coordinated it was," continues Shaul. "Was it 3 million bots or 10 million bots? We know it was a lot. We saw the use of all sorts of network protocols."

That's an under-statement. A bot comprising millions of compromised computers is a precious commodity in the world of cybercrime. Wielding it to take down one small blog site is aggressive to the point of overkill and also risks weakening it for future attacks as IP addresses and computers are spotted and closed down.

Either Krebs badly upset someone (the blogger has his own revenge theories as to the motivation) or something else was going on.

Shaul has a personal theory that it might simply be a malevolent marketing campaign designed to grab headlines in the right places.

"Why does someone want to spend this money and expose this massive botnet to us? Why is someone willing to expose themselves just because of something Bryan has said?

"My best guess is this is advertising for a new DDoS-for-hire service. It's a sort of launch for their service."

If that's true, the launch will have been noticed.

The attack raises several issues. First, DDoS attackers were able to 'persuade' Akamai to drop Krebs, who regularly publishes articles exposing the activities of Internet criminals, although he has since found a new home behind Google's Project Shield and is back online.

Akamai points out that it has quietly fended off regular DDoS attacks on Krebs pro bon but had to make a business decision that stopping the latest one was costing it money in mitigation and engineer time. It was a simple as that.

A second issue is that someone out there has a botnet large enough to cause Internet outages. The Spamhaus attacks was described at the time as 'slowing the Internet', a ridiculous description based on flimsy evidence. However, the Krebs DDoS does seem to have cause peering problems, the first time such a possibility has been publically acknowledged by the industry.

Perhaps it is in Akamai's interest to promote the attack's vastness but Shaul sounds sincere when he says, "we were literally watching the Internet struggle to cope with this attack."

It is now clear that someone out there has enough DDoS nous to shape traffic to cause real damage. It also has the volume and scale. The industry will adapt, says Shaul, but it had beter do that quickly.

"This attack has definitely moved the needle on how much capacity we need. It changes the model."

Show Comments