AS5617 TPNET

Networks observed carrying traffic for AS5617: the AS immediately before it in a routing path. Ranked by how many paths showed the relationship, which reflects how widely it is used, not how much traffic crosses it.

Inferred from path position, not from any declaration by either party. A network appearing here is carrying these routes; whether that is a paid transit arrangement, a backup, or something else is not something routing data can tell you.

Networks for which AS5617 was observed as the last hop before the origin, meaning it appears to provide them transit.

Every AS seen directly next to AS5617 in a routing path, in either direction.

Adjacency is not the same as peering. Being next to each other in a path covers paid transit, settlement-free peering and customer connections alike, and routing data cannot distinguish them. Treat this as "these networks exchange routes", nothing more specific.

Internet exchanges where Packet Clearing House has a port recorded for AS5617, largest fabric first. An exchange is shared switching where networks interconnect directly rather than paying a transit provider to carry traffic between them, so turning up at one is a deliberate choice about how this network wants to be reached. Each name opens that exchange's own page, with its peering LANs and every other network recorded on it.

No exchange membership on file. That is not the same as peering nowhere: the dataset records what PCH has collected, and many exchanges publish no participant list at all.

Unlike the tables above, this is recorded rather than inferred. An address on a peering LAN is at that exchange by definition. Two carve-outs all the same. Member ports counts every participant at the exchange, not this network's ports, so read it as a size proxy. And an absent exchange means PCH has no record of it, not that the network is not there.

Derived from routing tables collected at 2026-08-08 10:00 UTC by route-views2,rrc00, as seen from 75 BGP vantage points across 74.02M route entries. These are observations from those vantage points, not a complete or authoritative view of the internet.

Cross-check against RIPEstat or BGP.tools, which observe from a different and larger set of vantage points.

Exchange presence comes from a separate dataset: the directory collected 2026-08-08 by Packet Clearing House (https://www.pch.net/ixp/data), covering 1,327 exchanges and 93,137 recorded member ports, licensed CC BY-NC-SA 3.0.

PeeringDB is the better-known directory and is deliberately not the source here: its acceptable-use policy forbids the bulk storage this needs. The two answer different questions anyway. PeeringDB reports what an operator declares about itself; this reports what PCH has recorded.