Skip to content
INTERNET INFRASTRUCTURE & NETWORKS

Cloudflare Mitigates 18.2 Million DDoS Attacks in Q3 2026 as Threat Landscape Shifts Toward High-Volume Swarms

The global cybersecurity landscape experienced a massive wave of automated disruptions during the third quarter of 2026, according to the latest threat intelligence report released by Cloudflare. Returning to its standard quarterly cadence after releasing a combined volume analysis for the first half of the year, Cloudflare’s threat intelligence organization, Cloudforce One, documented an unprecedented frequency of distributed denial-of-service (DDoS) attempts. During July, August, and September, automated defense systems across Cloudflare’s global network autonomously mitigated an astonishing 18.2 million network-layer DDoS attacks alongside 14.4 trillion HTTP DDoS requests.

This staggering volume translates to roughly 8,226 network-layer attacks every single hour, or approximately 197,400 incidents per day. Put into a stark operational perspective, Cloudflare’s automated defenses intercepted a malicious network-layer attack once every 0.44 seconds throughout the quarter. In total, the mitigated traffic reached monumental proportions, scaling to 272.6 petabytes at the network layer and another 95.1 petabytes at the HTTP layer. These figures underscore the relentless nature of modern cyber threats and highlight the critical reliance organizations place on autonomous, real-time mitigation systems capable of neutralizing threats before human analysts can even review an alert.

More Swarms, Fewer Giants

A closer examination of the tactical trends driving Q3 2026 reveals a significant shift in how attackers deploy their resources. Rather than relying heavily on single, record-breaking monolithic floods, threat actors increasingly favored short-burst, mid-range, and high-count attacks. This tactical signature is characteristic of DDoS-for-hire platforms, commonly referred to as "booter" services. These low-cost, easily accessible subscription tools allow malicious actors to launch frequent, disruptive swarms of traffic against specific targets with minimal technical expertise.

The primary beneficiaries of this unwanted attention during the quarter were computer software companies, the gambling and casino sector, and broader information technology industries. The surge in these high-frequency, mid-range assaults drove the total count of network-layer attacks up by 208 percent compared to the same period in 2025.

Conversely, network-level hyper-volumetric attacks—defined as massive onslaughts exceeding one terabit per second—experienced a notable reversal. After surging dramatically to 805 incidents in the second quarter of 2026, the number of hyper-volumetric attacks dropped significantly to 251 during Q3. Security analysts attribute this cooling-off period directly to coordinated law enforcement actions earlier in the year that successfully dismantled four of the world’s largest IoT DDoS botnets: Aisuru, KimWolf, JackSkid, and Mossad.

Despite the disruption of those major operations, Internet of Things (IoT) botnets remain a durable and high-volume threat. The digital ecosystem continues to churn out resilient new variants, such as Dysphoria and Evooo1Bot, which continually infect insecure connected devices and routers to repurpose them as traffic relay nodes. Legacy threats also maintained a heavy footprint; Mirai-classified mitigations peaked in August alone at 57.23 petabytes. This single classification accounted for 46.5 percent of all mitigated network-layer bytes during the entire third quarter. The intensity was highlighted by four distinct days—August 16, 19, 23, and 25—where mitigated traffic cleared three petabytes each day. Furthermore, the longest sustained Mirai-related attack of the quarter ran continuously for nearly 50 hours between August 19 and August 21.

Vector Shifts: The Collapse of CLDAP and the Rise of UDP

The quarter also witnessed dramatic shifts in the specific protocols and techniques leveraged by attackers to amplify their payloads. The most pronounced change was a sudden and steep decline in Connectionless Lightweight Directory Access Protocol (CLDAP) reflection attacks. Earlier in the year, CLDAP attacks—where threat actors send small queries with forged victim IP addresses to public domain controllers over UDP port 389, tricking them into reflecting massive, amplified responses back at the target—experienced a 580 percent quarter-over-quarter surge. However, in Q3, this vector collapsed dramatically, accounting for a negligible 0.05 percent of all observed attacks.

Cloudflare DDoS Threat Report 2026 Q3: More swarms, fewer giants

At the same time, other reflection and volumetric vectors filled the void. DNS amplification attacks surged by 215 percent to account for 3.43 percent of all network-layer attacks. These operations exploit open DNS resolvers in a mechanism similar to CLDAP, using small spoofed queries to generate oversized replies directed at the victim.

More broadly, standard UDP floods climbed sharply from 14.06 percent to 31.31 percent of network-layer attacks. Because the User Datagram Protocol is connectionless and does not require a handshake before data is transmitted, it serves as an efficient vehicle for raw volumetric attacks. Malicious actors flood random or specific ports on a target system with UDP packets, forcing the host to constantly check for a listening application and respond, thereby exhausting its network bandwidth and computational resources.

Most-Attacked Industries and Geographic Surges

While the media, production, and publishing sector remained the single most-attacked industry throughout the third quarter of 2026, its overall share of malicious traffic dropped to roughly a quarter of what it experienced in Q2. As attackers diversified their targets, other sectors experienced unexpected surges. The internet industry climbed from outside the top ten into the sixth position, while consumer services and adult entertainment rose to eighth and ninth place, respectively.

Geographically, the distribution of targeted locations saw notable volatility. Canada experienced a dramatic climb from sixteenth to sixth place, driven largely by sustained attacks directed at its retail, telecommunications, service provider and carrier, and marketing and advertising industries amid a turbulent period of bilateral trade discussions. Meanwhile, Israel moved up seven spots to claim the tenth position, fueled by a wave of DDoS activity in mid-September. This defensive strain coincided with contentious political decisions ahead of parliamentary elections and the Rosh Hashanah holiday, alongside a broader, documented surge in hostile state-sponsored cyber warfare targeting the region throughout the year.

Globally, the United States retained its position as the most-attacked location in Q3 2026, with China following closely behind in second place. Indonesia also saw a notable rise, entering the top three with 2.93 percent of total attacks. This ongoing rotation between the United States and China at the top of the list mirrors trends from the first half of the year, when China held the top spot and the U.S. ranked second, reflecting shifting attacker preferences and geopolitical focal points.

Strengthening Global Defenses

Cloudflare officials emphasized that the network’s architecture is specifically designed to absorb this relentless year-over-year growth in DDoS threats. Every service operating on the network benefits from free, unmetered DDoS protection deployed across more than 330 cities globally, backed by a robust 500 terabits per second of network capacity. Because the vast majority of modern attacks conclude long before a human security analyst could identify an alert, full autonomy remains the cornerstone of effective mitigation.

To address the root cause of these network-layer assaults, Cloudflare continues to provide a free DDoS Botnet Threat Feed designed specifically for hosting providers, cloud computing platforms, and internet service providers. This initiative allows network operators to identify and dismantle abusive IP addresses and compromised accounts used to launch botnet swarms. More than 800 networks worldwide have already integrated the feed, fostering community-driven collaboration to neutralize threat infrastructure at its source.