Summary

  • RIPE associates "Pronet" LLC with four IPv4 allocations totalling 9,472 addresses, two IPv6 allocations and AS48467.
  • In RIPEstat’s two-week window ending at 08:00 UTC on 5 September 2026, one /24 timeline ended on 31 August while several component /24 timelines began on 2 September alongside covering /22 routes.

The RIPE NCC member entry lists “Pronet” LLC in Moscow and says it serves the Russian Federation. The organisation object identifies ORG-LA1704-RIPE as an LIR and records maintainer cwn-mnt. An inverse organisation search associates it with four ALLOCATED PA IPv4 blocks, two ALLOCATED-BY-RIR IPv6 blocks and AS48467.

That registry inventory is a useful perimeter, but it is not a routing snapshot. The AS48467 aut-num object names PRANET-AS and publishes import and export policy lines involving several networks. Those lines are declared policy. They do not establish realised traffic, contractual dependence or the physical paths in use.

RIPEstat’s routing-status view gives a bounded observation at 08:00 UTC on 5 September: ten IPv4 prefixes covering 2,304 addresses, one IPv6 prefix covering 65,536 /48s, 19 observed neighbours and visibility through all RIS peers listed in the response. None of those numbers is a service-level measure. They do not prove global reachability, utilisation, customer scale or resilience.

The more revealing record is the announced-prefix timeline. Its window runs from 22 August to 5 September. The entry for 94.228.167.0/24 ends on 31 August. The covering 94.228.172.0/22 appears through the window, while 94.228.173.0/24, 94.228.174.0/24 and 94.228.175.0/24 appear from 2 September. A similar pattern is visible under 185.46.44.0/22: the covering route spans the window and all four component /24s appear from 2 September.

That sequence does not disclose why the routes changed. Aggregate and more-specific announcements can coexist for many reasons, and the public sources do not identify an incident, migration, customer arrangement or traffic-engineering decision. The evidence supports a narrower conclusion: a single count hides movement. Diligence should keep the registry inventory, declared policy, current observed footprint and change timeline as separate layers.