ProbeNet is not our only data source, but it is the foundation that lets us “trust but verify” every dataset we build.
At IPinfo, our ProbeNet actively measures the global internet by running internet measurements of 100% of routed IPv4 addresses (~3.12B) and about 3.7B IPv6 addresses that we identify as actively in use. These probes use techniques like ping, traceroute, and other tests to provide direct, evidence-based insight into IP behavior.
However, not every address responds. Firewalls, NAT, and provider policies often block measurement, and many networks intentionally configure addresses not to reply. In practice, around 610 million IPs respond to measurement, approximately 481 million IPv4 and 129 million IPv6. That represents about 15–20% of IPv4 space and just 3–4% of IPv6.
This distinction is important:
- Probed IPs = all the addresses we attempt to measure.
- Pingable IPs = the subset that actually respond.
ProbeNet’s uniqueness lies in providing “ground truth” where responses exist, which we then combine with other data sources and inference techniques to cover the remainder. For example, neighboring IPs in the same range (like /24 for example) often share similar properties, allowing us to extend measurements beyond direct responders.
While only a fraction of the internet is directly pingable, running internet internet measurements the entire address space is essential. It provides the hard evidence that third party data and registry (WHOIS, geofeed) information alone cannot deliver. This evidence-based approach ensures that IPinfo data is both more accurate and more defensible than relying on third-party sources in isolation.