5.8 Million Knocks: Countries and Networks of Origin

Report: 5.8 Million Knocks: A Three-Year Honeypot Record of SSH and Telnet Attacks

Attacks come mainly from cloud and hosting networks, while scans come mainly from national carriers. China, the United States and the Netherlands lead the country rankings for both scans and attacks. Chinanet, China Unicom and Pfcloud lead the network rankings for scans, while DigitalOcean, Tencent and TechTies lead the rankings for attacks. Each share below is measured against the total shown at the head of its column.

Rank All traffic (5,810,156) Scans (1,247,416) Attacks (4,562,740)
1 China 17.5% China 35.6% United States 13.2%
2 United States 13.0% United States 12.4% China 12.5%
3 Netherlands 10.3% Netherlands 8.7% Netherlands 10.7%
4 Singapore 5.1% India 5.7% Singapore 5.9%
5 India 4.7% Germany 5.0% India 4.5%
6 Germany 4.4% Pakistan 3.7% Germany 4.2%
7 Bulgaria 3.4% Iran 3.3% Bulgaria 4.1%
8 Russia 3.0% Russia 2.2% Russia 3.1%
9 South Korea 2.6% Singapore 2.1% Indonesia 3.1%
10 Hong Kong 2.6% South Korea 1.9% Hong Kong 2.9%
Table 1. Unsolicited traffic by country
Rank All traffic (5,810,156) Scans (1,247,416) Attacks (4,562,740)
1 AS14061 DigitalOcean 12.0% AS4134 Chinanet 15.2% AS14061 DigitalOcean 13.8%
2 AS4134 Chinanet 5.3% AS4837 China Unicom 7.7% AS132203 Tencent 6.1%
3 AS132203 Tencent 4.8% AS51396 Pfcloud 5.9% AS197170 TechTies 6.0%
4 AS197170 TechTies 4.7% AS14061 DigitalOcean 5.8% AS45090 Tencent 2.8%
5 AS51396 Pfcloud 2.9% AS9829 BSNL 3.4% AS4134 Chinanet 2.6%
6 AS45090 Tencent 2.6% AS58224 Iran Telecom 3.1% AS51396 Pfcloud 2.1%
7 AS4837 China Unicom 2.0% AS396982 Google Cloud 2.7% AS198364 BanatSync 2.1%
8 No ASN mapped 1.9% AS197071 Active 1 2.5% No ASN mapped 1.9%
9 AS198364 BanatSync 1.6% AS45090 Tencent 2.0% AS16276 OVH 1.7%
10 AS4766 KT Corporation 1.5% AS9541 Cyber Internet Services 1.8% AS47890 Unmanaged Ltd 1.6%
Table 2. Unsolicited traffic by autonomous system

What the rankings and their underlying data show:

  1. Chinese scanning runs through national carriers. China accounts for 35.6% of all scan traffic, nearly three times the United States (12.4%). Its two backbone carriers, AS4134 and AS4837, hold 22.9% of all detected scans.

  2. Attacks run through cloud and hosting providers. DigitalOcean (AS14061) accounts for 13.8% of global attack traffic, more than twice the next network. DigitalOcean operates in several of the countries on the list, which plausibly explains part of the traffic attributed to Singapore, the Netherlands, India and Germany. The two Tencent ASNs and the Chinanet ASN combined account for 11.5%.

  3. The networks are concentrated. IP01 (see [IP analysis post for details])(/blog/index.php/2026/11/11/5-8-million-knocks-busiest-ips-as-origin-of-attacks) made 66,742 scans and 84,295 attacks from Pfcloud (AS51396), amounting to 5.4% of all scans and 1.8% of all attacks. The ten leading scan networks carry 50% of all scans (624,261 of 1,247,416), and the ten leading attack networks carry 40.7% of all attacks (1,856,899 of 4,562,740). By contrast, the ten busiest IPs account for only 12.7% of scans and 7.9% of attacks, so network-level controls reach much further than IP-level ones.

  4. The Netherlands is third everywhere. It holds 8.7% of all scans and 10.8% of all attacks. One Dutch-labelled address, IP01, made 5.4% of all scans and 1.8% of all attacks.

  5. City-scope jurisdictions made the top list. Singapore is a particularly relevant case: although the country operates 0.75% of all IPv4 addresses, the honeypot detected that 2.1% of all scan traffic and 5.9% of all attack traffic originated there. Hong Kong shows a similar pattern, with 2.9% of all attacks coming from IPs operating there.

  6. Scan and attack networks are largely separate. AS132203, AS197170, AS198364, OVH (AS16276) and Unmanaged Ltd (AS47890) appear only on the attack side, whereas BSNL (AS9829), Iran Telecom (AS58224), Google Cloud (AS396982), Active 1 GmbH (AS197071) and Cyber Internet Services (AS9541) appear only on the scan side. The Google Cloud share may include benign research scanners, which the data cannot separate.

  7. Traffic with no attributed ASN was detected. At the time it was consulted, the MaxMind GeoLite2 database failed to attribute an ASN to a number of IP addresses responsible for a significant volume of attacks: 86,481 of 4,562,740, or 1.9%.

Provider-level reporting and rate limiting would reach furthest, since a handful of networks carry a large share of the volume. The cost of network-wide blocks is collateral damage to legitimate users on the same networks, which cannot be quantified from this data.

Report: 5.8 Million Knocks: A Three-Year Honeypot Record of SSH and Telnet Attacks


Comments

Leave a Reply

Your email address will not be published. Required fields are marked *