Probes

A probe is a small device hosted by a volunteer on their own network, from which RIPE Atlas runs measurements. Each has a stable numeric ID, one or two address families, a location, an ASN and prefix per family, a firmware version and a set of tags.

The archive is the point

RIPE's API tells you what a probe looks like now. Atlas additionally imports a full snapshot of every probe RIPE has on file — around 50,000, not just the connected ones — once per day, and keeps every snapshot.

That daily archive is what makes historical questions answerable: what ASN was this probe in last March, when did it move country, how many probes did this network have a year ago. Anything you see in a Trends or Deployments view over a long window is ultimately reading it.

The trade-off is resolution: the archive is daily. A probe that changed ASN and changed back within a day leaves no trace in it. For sub-day resolution on a single probe, the connection events and uptime change-points are the finer-grained record.

Status

A probe moves through a small set of states:

• Never connected — registered but never came up
• Connected — reporting normally
• Disconnected — not currently reporting
• Abandoned — down long enough that nobody expects it back

Atlas records not just the current status but how long it has been in it, which is usually the more useful number: "Disconnected" says little, "Disconnected for 43 days" says a lot.

Identity and drift

Probes are volunteer hardware on volunteer networks, and they drift:

• ASN and prefix change when the host switches provider, and are tracked separately for IPv4 and IPv6 — a probe can be in different ASNs per family.
• Source IP changes on renumbering, which shows up as an uptime change-point.
• Firmware updates over time, also a change-point.
• Location is what the host declared. It is not verified, and it is frequently approximate. See Geofeeds and geolocation for what Atlas does about that.
• Tags are partly assigned by RIPE and partly by the host, so treat them as hints rather than facts.

Health

Beyond up/down, Atlas evaluates probes against a set of rules and records the resulting issues, including whether a probe is actually producing usable results for the measurements it participates in — a probe can be connected and still be contributing nothing.

Probe health and probe result health are staff views.

Anchors are probes too

Every anchor is also a probe and appears in probe data. What makes it an anchor is the additional role of being a measurement target. See Anchors.

See the Probes guide for the pages built on all this.