惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

U
Unit 42
罗磊的独立博客
博客园 - 聂微东
T
The Blog of Author Tim Ferriss
博客园 - 司徒正美
Stack Overflow Blog
Stack Overflow Blog
F
Fortinet All Blogs
A
About on SuperTechFans
腾讯CDC
Apple Machine Learning Research
Apple Machine Learning Research
B
Blog RSS Feed
IT之家
IT之家
V
Visual Studio Blog
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
宝玉的分享
宝玉的分享
C
Check Point Blog
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
Vercel News
Vercel News
爱范儿
爱范儿
Microsoft Security Blog
Microsoft Security Blog
月光博客
月光博客
T
Tailwind CSS Blog
The Cloudflare Blog
Hugging Face - Blog
Hugging Face - Blog

A10 Networks

The Evolution of AI Red Teaming: Lessons Learned | A10 Networks How to Secure AI Applications From Modern Threats? | A10 Networks How to Secure AI Applications From Modern Threats? | A10 Networks A10 Thunder 3360S ADC: 150 Gbps for AI & Enterprise | A10 Networks AI Security Is an Architecture Problem | A10 Networks What Is an AI Gateway? The AI Control Plane | A10 Networks AI in Financial Services: Why Leadership Is the Risk OWASP Now Has Two AI Top 10s: What Changed? OWASP Now Has Two AI Top 10s: What Changed? TrojAI by A10 Networks AI Security Platform TrojAI Defend for MCP by A10 Networks TrojAI Detect by A10 Networks TrojAI Defend by A10 Networks Secure, High-Performance Networking Solutions | A10 Battling Bots, Fraud & AI Threats Summit | Retail IT & Cybersecurity What Is Healthcare Data Compliance? | A10 Networks Interop Best of Show Runner's Up - People's Choice | A10 Networks Interop Best of Show Runner's Up - Security for AI | A10 Networks What Is FIX Protocol Trading? | A10 Networks A10 Joins OpenAI's Trusted Access for Cyber Flexible Licensing for Multiple Clouds | A10 Networks A10 Acquires TrojAI to Advance Enterprise AI Security HFT Infrastructure: High Frequency Trading Explained | A10 Networks A10 Networks Acquires TrojAI Inc., Expanding AI Roadmap | A10 Networks What Is Low-latency Trading? | A10 Networks Multi-Vector DDoS: 11 Amplification Vectors | A10 Healthcare Cloud Compliance: HIPAA & GDPR Guide | A10 LLM Unbounded Consumption & DoS Attacks | OWASP LLM10 LLM Hallucination & Misinformation | OWASP LLM09:2025 Healthcare Network Protection for Hospitals & Clinics
Real-time DDoS Carpet-bombing: NTP Amplification Evasion
Anidhya Pand · 2026-05-05 · via A10 Networks

DDoS Carpet-bombing Attack Unfolds in Real Time. Here’s What It Looked Like.

25 Ips, one subnet, 40 minutes, and your per-IP rate limiter didn’t see a thing

A Quiet Morning in South Asia

At 07:35 UTC on April 28, 2026, a regional ISP in South Asia started taking fire. Not the kind of fire that trips alarms. No single IP experienced a traffic spike dramatic enough to cross a typical per-IP detection threshold. But something far more deliberate was underway.

Over the next 40 minutes, we watched through A10 Defend Threat Control as attackers systematically walked through more than 25 distinct IP addresses within a single /24 subnet, hitting each one with high-complexity NTP amplification. A new target every 30 to 90 seconds. Each attack lasted anywhere from 21 seconds to nearly 8 minutes. Then the attacker moved on to the next address.

This is carpet-bombing and it is one of the most effective DDoS evasion techniques in use today.

What Carpet-bombing Looks Like in the Data

Here is a condensed view of the attack timeline, pulled directly from the A10 Defend Threat Control victim data. Every row is a separate attack event. Every IP belongs to the same /24 owned by the same regional broadband provider.

Time (UTC)Victim IP (Masked)DurationVector
07:35:17160.250.82.x (.45)21 secNTP – High
07:35:41160.250.82.x (.242)404 secNTP – High
07:36:16160.250.82.x (.115)57 secNTP – High
07:37:14160.250.82.x (.56)80 secNTP – High
07:38:36160.250.82.x (.207)54 secNTP – High
07:39:32160.250.82.x (.43)63 secNTP – High
07:40:40160.250.82.x (.232)37 secNTP – High
07:42:27160.250.82.x (.176)49 secNTP – High
07:43:20160.250.82.x (.141)354 secNTP – High
07:44:26160.250.82.x (.70)56 secNTP – High
07:45:25160.250.82.x (.215)77 secNTP – High
07:46:44160.250.82.x (.254)29 secNTP – High
07:47:15160.250.82.x (.229)67 secNTP – High
07:49:16160.250.82.x (.85)88 secNTP – High
07:50:51160.250.82.x (.117)37 secNTP – High
… (25+ IPs total)
08:01:14160.250.82.x (.131)479 secNTP – High

Figure: Condensed timeline of the carpet-bombing campaign. Full dataset available in A10 Defend Threat Control.

Notice the pattern. No single IP accumulates enough traffic to dominate a dashboard. The attacker distributes the pain across the entire subnet, ensuring that the aggregate bandwidth hitting the victim’s upstream link is enormous, even though each individual target looks only moderately busy.

Why This Matters More Than You Think

Carpet-bombing exploits a fundamental assumption baked into most DDoS detection systems: that attacks concentrate on a single destination. Traditional mitigation triggers on per-IP volumetric thresholds. When traffic to a single IP spikes, you scrub it. Simple.

But what happens when no single IP crosses the threshold? When the attacker instead sends 40 Gbps spread across 25 IPs, each absorbing just 1.6 Gbps. Individually, those flows might sit below your alerting baseline. Collectively, they saturate the upstream link and take the entire subnet offline.

This is not theoretical. In this campaign, the targeted ISP’s entire /24 was under sustained NTP amplification for 40 continuous minutes. The longest single-IP attack lasted nearly 8 minutes (479 seconds), but the campaign lasted far longer because the attacker simply rotated targets.

The Clues Were There—If You Knew Where to Look

What made this campaign visible in A10 Defend Threat Control was not any single attack event, but the correlation across events. The platform captures victim data across all observed attacks globally, which means patterns that span multiple IPs, subnets, and time windows become visible in a way they never would from a single vantage point.

When we queried the data for this /24, the carpet-bombing pattern jumped out immediately: same ASN, same attack vector, same complexity rating, tightly clustered timestamps, and sequential IP targeting. An analyst looking at any individual event would see a routine 60-second NTP hit. An analyst looking at the subnet-level pattern would see an orchestrated campaign.

That difference—between seeing one event and seeing the campaign—is the difference between reacting and anticipating.

“An analyst looking at one event sees a routine NTP hit. An analyst looking at the subnet sees an orchestrated campaign. That gap is where carpet-bombing lives.”

What Should Defenders Do?

  • Think in subnets, not just IPs. If your detection logic only evaluates per-destination thresholds, carpet-bombing will fly under the radar. Aggregate traffic analysis at the /24 or prefix level is essential.
  • Correlate across time windows. A 60-second NTP attack is unremarkable. Twenty-five of them against the same subnet in 40 minutes is a campaign. Your tooling needs to surface that pattern automatically.
  • Use external threat intelligence. The victim ISP in this case may not have had visibility into the broader pattern. They only see their own ingress. A platform like A10 Defend Threat Control provides the global vantage point that turns isolated alerts into actionable intelligence.
  • Assume NTP amplification isn’t going away. Every attack in this campaign used NTP reflection at high complexity. Open NTP servers continue to provide massive amplification, and attackers know it.

See it for Yourself

This carpet-bombing campaign is just one pattern we surfaced from over one million DDoS attacks observed by A10 Defend Threat Control in April 2026 alone. The platform lets you filter by attack type, victim country, ASN, industry, duration, and more, turning raw data into the kind of intelligence that keeps you ahead of the next campaign.

Explore the live data on the A10 Defend Threat Control dashboard or sign up for a 90-day free trial to see what attackers are doing right now.



Anidhya Pandita

|

May 4, 2026