bgpreport

Network · data from 2026-11-12

AS20647 IPB Internet Provider Berlin GmbH

Registry name: IPB IPB Internet Provider in Berlin GmbH, DE · PeeringDB

8IPv4 routes 62.5% RPKI valid · 100.0% with a route object
3IPv6 routes 66.7% RPKI valid · 66.7% with a route object
0RPKI invalid 0 origin · 0 maxLength
4No ROA 1 with no route object for this origin
0Low visibility seen by under 10% of full-table peers
0Route objects not announced registered with origin AS20647, seen from no peer

Announced routes

Every prefix a peer of the route collectors sees with origin AS20647, things to look at first. Visibility = full-table peers that see it, of 673 (IPv4) and 636 (IPv6) on 75 collectors. Also as routes.csv.

2001:67c:2818::/48 No ROA Other origin only 632 99% 68
91.233.84.0/22 No ROA RIR databaseRIPE 673 100% 66
193.178.163.0/24 No ROA RIR databaseRADB, RIPE 673 100% 66
195.20.159.0/24 No ROA RIR databaseRIPE 673 100% 66
2.57.160.0/22 Valid RIR databaseRIPE 673 100% 66
91.102.8.0/21 Valid RIR databaseRIPE 673 100% 66
185.27.156.0/22 Valid RIR databaseRIPE 673 100% 66
185.231.124.0/22 Valid RIR databaseRIPE 673 100% 66
194.29.224.0/19 Valid RIR databaseRIPE 673 100% 66
2a02:f28::/32 Valid RIR databaseRIPE 636 100% 69
2a0c:5f00::/32 Valid RIR databaseRIPE 636 100% 69

AS-SET

PeeringDB's irr_as_set for AS20647: AS-IPB. Expanded from the day's IRR dumps the way bgpq4 does against whois.radb.net: a name with a database is read from that database only, one without from the first database in RADB's order. "Not covered" = this network's own routes that a filter built from the set would leave out (AS20647 isn't in the set, or no route object for AS20647 covers the route; more-specifics of a route object are accepted down to /24 and /48, as with bgpq4 -A -R 24 / -R 48). Customers' routes aren't counted: the collectors show each route's origin, not which network it came through.

NameDatabaseASNsSets Route objectsAS20647 in itNot covered
AS-IPB RIPE 379351 Yes 1

Fix list

The changes that would bring RPKI, the IRR and what's announced into line. Some are deliberate (a route object kept for a backup route, say); the network decides. All 5 are in fix.csv.

  • ROA: none4
  • Route object: none1
ChangePrefix or setDetailFull-table peers
ROA: none2001:67c:2818::/48No ROA covers it. Add a ROA for AS20647, maxLength 48.632 of 636
Route object: none2001:67c:2818::/48Route objects for it name AS198703. Add a route object with origin AS20647 in the RIR database for the space.632 of 636
ROA: none91.233.84.0/22No ROA covers it. Add a ROA for AS20647, maxLength 22.673 of 673
ROA: none193.178.163.0/24No ROA covers it. Add a ROA for AS20647, maxLength 24.673 of 673
ROA: none195.20.159.0/24No ROA covers it. Add a ROA for AS20647, maxLength 24.673 of 673

Route objects not announced

Every route object with origin AS20647 is announced.

Routes: the 08:00 UTC snapshots of 2026-11-12 from 75 RIPE RIS and RouteViews collectors. RPKI: our rpki-client run before the snapshot. IRR: the day's dumps of the five RIR databases and RADB's mirrors. The LEVEL3 and IDNIC databases publish no dump and aren't read, so route objects only there count as none.