Probe the observed routing policy for an IPv4/IPv6 address, a hostname, or an IPv4 prefix
The Internet2 measurement system holds three probe source addresses.
Their prefixes are announced twice — once into the R&E fabric
(Internet2, AS11537) and once to the commodity Internet (AS396955)
— with a deliberate asymmetry for each source. Every test pings
the target from all three sources; the target's network routes each
reply back toward the source using its own BGP best-path selection.
Because the two announcements differ, the side each reply arrives on
(I2_re vs I2_commodity) reveals one specific
property of the target's routing policy. The diagram shown with
your results depicts that measurement: each colored
arrow shows the side that test's reply actually came back on.
For an IPv4 prefix, up to ten random addresses with a recent
history of answering pings (USC/ISI responsive-address data) are
tested independently, and the report shows up to five of them
— preferring the addresses that replied.
| Test | Announcement asymmetry | Reading the reply |
|---|---|---|
R&E preference163.253.63.632001:468:2f00::1 |
Announced on both sides, but 4× AS-path prepended toward R&E — the R&E path is deliberately the longer one. | Reply via I2_re: the network overrides the longer path — an explicit Local Preference for R&E. Reply via I2_commodity: it simply followed the shorter path (no R&E preference shown). |
Commodity preference163.253.64.12001:468:2f01::1 |
The mirror image: announced on both sides, 4× prepended toward commodity — the commodity path is deliberately the longer one. | Reply via I2_commodity: an explicit preference for commodity despite the longer path. Reply via I2_re: it followed the shorter path. |
Full table163.253.67.12001:468:2f03::1 |
Split by prefix length: a more-specific (/24, /48) announced only via commodity; only a covering route (/23, /47) via Internet2. |
Reply via I2_commodity: the network carries the commodity more-specific — a full table. Reply via I2_re: it doesn't — a partial commodity table (or default) that falls back to the R&E covering route. |
| Verdict | Pattern | Interpretation |
|---|---|---|
| R&E local preference | reply via I2_re, reply via I2_commodity |
Explicit Local Preference for R&E paths, and the full-table test confirms a complete commodity table. The ideal R&E configuration. |
| Default toward R&E | and both reply via I2_re |
This network doesn't appear to carry a full commodity routing table. The full-table test's address is announced most specifically to commodity, yet the reply still came back via R&E — meaning the network fell back to its R&E routes. |
| No local-preference override | and reply on different sides | Each reply follows whichever announcement is shorter — R&E and commodity have equal Local Preference and AS-path length decides. |
| Commodity preference | and both reply via I2_commodity |
Replies use commodity even where the commodity announcement is 4× prepended — an explicit preference for commodity paths over R&E. |
| No replies | None of the three tests got a reply | The address answered pings in the recent past (that's why it was selected) but appears not to any more — its routing policy could not be observed. |
| Inconclusive | A mixed pattern | Some replies returned, but the pattern does not match a known classification. |
Sharing: after a probe runs, the page URL ends in
#<target> — copy it to share this test; opening
the link re-runs the probe for that target.