Bitcoin was one of the most targeted industries
Amidst the cryptocurrency price spike, bitcoin was one of the top-10 most targeted industries, despite its relatively small size.
High packet rate attacks grew more common
144 attacks (2.9 percent) went above the 100 Mpps mark, compared to just six in the first quarter of the year.
A third of network layer attacks were highly persistent
29.6 percent of network layer targets were hit ten or more times, nearly triple the number of application layer targets.
Botnet activity out of India and Turkey continued to climb
Following an increase in Q2, botnet activity out of the two countries continued to climb, reaching 11.2 percent.
Q3 2017 was marked by an improvement in on our sampling methods and methodology, which changed how we define a single distributed denial of service (DDoS) attack in our quarterly reports.
In prior reports, a DDoS event was defined as a single assault that was preceded by a quiet (attack free) period of at least ten minutes, and followed by another quiet period of the same duration or longer. This quarter, however, we increased the quiet period to sixty minutes in order to aggregate successive attacks.
This was done in response to an increase in the number of short-lived repeat DDoS attacks, such as hit-and-run and pulse wave assaults. Our new sampling method puts these attacks in their proper context, observing them not as a series of independent incidents, but as a single persistent event.
In addition, we expanded on our research to provide additional data and insights. As a result, the scope of our quarterly report has more than doubled from nine to 19 data sets, enabling us to share a more in-depth and detailed view of the DDoS threat landscape.
In Q3 2017, we saw high packet rate attacks—assaults in which the packet forwarding rate escalates above 50 million packets per second (Mpps)—become more common. So much so that five percent of all network layer assaults came in above 50 Mpps, with the largest attack peaking at 238 Mpps.
144 attacks (2.9 percent) went above the 100 Mpps mark, compared to just six in the first quarter of the year. This is a cause for concern, as many mitigation solutions are ill equipped to process packets at such a high rate.
Hong Kong topped our list of the most targeted country for network layer assaults in Q3 2017, largely because of a persistent attack on a local hosting service that was hit hundreds of times throughout the quarter. The largest application layer assault targeted a financial services company headquartered in Europe, which was hit multiple times with attacks above 100,000 RPS.
These two attacks illustrate the ongoing macro trend of increased DDoS attack persistence. In Q3 2017, nearly a third of all organizations targeted by network layer assaults were hit more than ten times, as was the case for 11 percent of the domains targeted by application layer assaults.
Target wise, we witnessed a high number of attacks against the bitcoin industry in Q3 2017, which drew 3.6 percent of assaults despite its relatively small footprint.
Currently there are under seventy active bitcoin exchanges worldwide, less than half of which see over 10 million USD in daily trending volumes. In comparison, the retail industry, a much larger category that includes all e-commerce sites, drew 5.8 percent of DDoS attacks.
Finally, botnet activity out of Turkey and India continued to increase for the second quarter in a row. In Q3 2017, these two countries, which typically aren’t part of the top-10 attacking countries list, accounted for over ten percent of all botnet activity.
NETWORK LAYER ATTACKS
In Q3 2017, 8.6 percent of network layer attacks came in above 50 Gbps and five percent were above 50 Mpps. Out of these, 144 attacks (2.9 percent) reached above the 100 Mpps mark.
The largest attack of the quarter peaked at 299 Gbps and targeted Incapsula’s own IP ranges, a common occurrence as our IP masking services hide actual customer IP addresses behind our own. The highest attack rate, recorded during an assault targeting a forex company in Asia, came in at 238 Mpps, up from 190 Mpps in Q2 2017.
In contrast, the majority of attacks (90.2 percent) were under 10 Mpps and were predominantly the result of DDoS-for-hire activity.
TOP ATTACKED COUNTRIES
Despite being home to only 5.1 percent of targets, Hong Kong was targeted by almost a third of all network layer attacks in Q3 2017. This was largely due to a large-scale campaign against a local hosting service provider, which was hit more than 700 times throughout the quarter.
Taiwan and the Philippines made an atypical appearance on the top-10 list of attacked countries, following a number of large campaigns targeting gambling websites in their respective regions.
TOP ATTACKED INDUSTRIES
In Q3 2017, over a third of network layer attacks targeted online gambling sites and related services. This is common, as these sites are reliant on web revenue and exposed to extortion attempts. Additionally, they’re highly competitive and are commonly targeted by rival companies.
The online gaming industry was also frequently targeted, although in this case the attacks typically originate from users either attempting to influence a game’s outcome or just to vent their frustrations.
The high number of attacks that targeted the internet services industry was driven by the large campaign in Hong Kong. Even if this campaign were to be disregarded, however, internet services would still hold the first place as the most attacked industry.
This is because each of the businesses in this category, such as hosters and ISPs, individually represent hundreds or thousands of domains that use their services. As such, each represents a large attack surface that can draw multiple attacks from different offenders preying on various targets.
Lastly, in Q3 2017, we saw attacks targeting a relatively high number of cryptocurrency exchanges and services. This was likely related to a recent spike in the price of bitcoin, which more than doubled in the span of the quarter. As a result, bitcoin made the top-10 most targeted industries list, despite its relatively small size and web presence.
This young and exponentially growing industry presents a lucrative opportunity for extortionists and other cybercriminals who are always on the lookout for potentially vulnerable and high-profit targets.
In this specific case, the attacks could have been also launched to manipulate bitcoin prices, something offenders have been known to do.
In Q3 2017, our methodology for measuring attack duration was changed to provide a more accurate picture of the current threat landscape. Whereas previously short attacks were counted individually, repeat assaults taking place in the span of an hour are now treated as a single event. For example, three attacks spaced ten minutes apart are now recorded as one assault.
As a result, the number of attacks lasting more than six hours this quarter increased dramatically to 7.5 percent, from 0.8 percent in Q2 2017. The longest attack of the quarter lasted more than 5.5 days, while average attack duration was 1.2 hours.
In Q3 2017, half of network layer targets were hit at least twice, while almost 30 percent were attacked more than ten times. Considering the changes to our methodology, this means that nearly a third of targets were attacked ten or more times, with at least an hour interval in between assaults.
With an average attack duration of 1.2 hours, this also means that—on average—organizations targeted by DDoS offenders spent 12 hours under attack over the course of the quarter.
In Q3, we saw a steep increase in the number of amplification attacks. DNS-amplification assaults tripled from five percent in Q2 to 15.9 percent this quarter, while NTP-amplification attacks shot up to 36.9 percent from 9.9 percent.
In non-amplified assaults, we continued to see attackers use a variety of fabricated payloads. There was a clear preference towards SYN, TCP and UDP floods, which were often the three payload types used in the course of a single attack.
In Q3 2017, the number of multi-vector attacks increased to 70.2 percent, from 21.7 percent in the previous quarter. The dramatic increase is closely linked to the changes in how we measure attacks—now, an assault in which an attacker sends out rapid bursts of traffic (e.g., a pulse wave) using different packet types is considered a multi-vector attack.
Measuring attacks in this manner paints a more accurate picture of how sophisticated DDoS assaults have become, and how easily attackers can switch vectors on-the-fly and mix between amplified and non-amplified assaults.
APPLICATION LAYER ATTACKS
In Q3 2017, one in five application layer assaults went above 1,000 requests per second (RPS). The largest attack was against a financial services company hosted in Europe and clocked in at 134,486 RPS. The site was hit multiple times with attacks above 100,000 RPS, many of which, including the largest one, targeted the organization’s API gateways.
TOP TARGETING AND ATTACKING COUNTRIES
In Q3 2017, the US topped our list of the most attacked countries, both in terms of hosted targets (40.1 percent) and attacks (53.3 percent). Coming in second place, the Netherlands was home to 10.6 percent of targets and 8.8 percent of attacks.
The remainder of the list consisted of developed countries with mature digital marketplaces, including Singapore, Japan and Australia, which make for attractive targets.
The change in our methodology for measuring attack duration, (i.e., aggregating smaller assaults into a single event), allows us to disregard individual assaults that are part of a larger attack.
As a result, the number of attacks under 30 minutes fell from 57.4 percent in the previous quarter to 17.1 percent in Q3 2017. At the same time, attacks lasting between 30 minutes and six hours came in at 73.2 percent in Q3 2017, compared to 35.2 percent in the previous quarter.
The change in how we measure DDoS attacks gives us a more accurate view of attack persistence that disregards the “noise” made by short-lived assaults occurring in rapid succession.
As a result, we saw the number of repeat attacks drop from 75.8 percent in the previous quarter to 46.7 percent in Q3 2017. Even with the more precise measurement, however, we still saw that almost 16 percent of targets were exposed to six or more attacks throughout the quarter.
In Q3 2017, nearly 17 percent of botnet traffic originated in China, despite the fact that it hosted over 40 percent of attack devices. This represented a significant drop from the previous quarter, when China was the source of 63 percent of bot traffic.
After a notable uptick in Q2 2017, botnet activity in Turkey and India continued to increase this quarter. A staggering 7.2 percent of botnet traffic originated in Turkey in Q3 2017, up from 2.1 percent in the previous quarter. In India, that figure increased to four percent from 1.8 percent in the prior quarter.
These countries, which have rarely appeared in the top-10 attacking country list thus far, accounted for over 10 percent of all botnet activity while serving as a home to 5.1 percent of all attacking devices.
Our analysis is based on data from 3,920 network layer and 1,755 application layer DDoS attacks on websites using Imperva Incapsula services from July 1, 2017, through September 30, 2017—referred to herein as the third quarter of 2017 or Q3 2017.
Information about DDoS bot capabilities and assumed identities comes from a random sample of 37.4 billion DDoS attack requests collected from such assaults over the same period.
DDoS attack – A persistent, distributed denial of service event against the same target (e.g., IP address or domain). A single attack is preceded by a quiet (attack free) period of at least a sixty minutes, and followed by another quiet period of the same duration or longer.
Network layer attack – An assault against either the network or transport layers (OSI layers 3 and 4). Its goal is to cause network saturation by expending much of the available bandwidth. It’s typically measured in gigabits per second (Gbps), referring to the amount of bandwidth it can consume per second.
Application layer attack – An assault occurring on OSI layer 7. Its goal is to bring down a server by exhausting its processing resources (e.g., CPU or RAM) with a high number of requests. It’s measured in requests per second (RPS)—the number of processing tasks initiated per second. Such attacks are executed by DDoS bots able to establish a TCP handshake to interact with a targeted application.
Botnet – A cluster of compromised, malware-infected devices remotely controlled by an offender. Device owners are unaware of their system participation.
DDoS bot – A malicious software application (script) used by a perpetrator. So-called bad bots only come into play in application layer attacks, where a TCP connection is established. They typically masquerade as browsers (human visitors) or legitimate bots (e.g., search engine crawlers) to bypass security solutions
Payload – In the context of this study, a payload is a packet type used in a network layer assault. It’s fabricated by an attack script and can often be altered on the fly. In many cases, multiple payload types are used simultaneously during the course of a single event.