[ITNOG] FW: ARIN removed 154 as-set's from it's IRR a week+ ago

[Brian Turnbow]

Questo da Nanog, 
non so se qualcuno usa IRR di ARIN,  ma è info utili.

Brian

Brian Turnbow
CTO 



TWT S.p.A.

> -----Original Message-----
> From: NANOG [mailto:nanog-bounces+b.turnbow=twt.it@nanog.org] On
> Behalf Of Martin J. Levy
> Sent: Thursday, June 25, 2020 3:37 AM
> To: nanog@nanog.org
> Subject: ARIN removed 154 as-set's from it's IRR a week+ ago
> 
> This is a quick note about as-set's.
> 
> First some background. For those that followed the ARIN to ARIN & ARIN-
> NONAUTH split a few weeks ago, you'll know that ARIN moved around 56% of
> all their IRR route/route6 objects into the ARIN-NONAUTH source database.
> This was based on objects that are not owned within ARIN managed space.
> For more on this; review the ARIN mailing lists and the video from their
> meeting a few weeks ago.
> 
> But what about all those as-set objects within ARIN IRR? This wasn't really
> reviewed heavily and so I thought, I'd dump some info everyone's-way. Yes
> there are objects and yes, some moved to ARIN-NONAUTH (and hence maybe
> aren't being used for route filtering anymore).
> 
> First off - this data is based on the ARIN IRR database dump of 24/June/2020.
> You can find the current IRR data at ftp://ftp.arin.net/pub/rr/ - it's easy to
> download the dataset. Here's the raw numbers:
> 
> $ gunzip < arin-nonauth.db.gz | egrep -i '^as-set:' | wc -l
> 154
> $ gunzip < arin.db.gz | egrep -i '^as-set:' | wc -l
> 1413
> $
> 
> Only around 154 as-set objects were placed into ARIN-NONAUTH. Around
> 9.8% of the total number of objects.
> 
> Running a quick script against the RADB combined IRR database shows that of
> those 154 as-set objects that ARIN moved; there're 133 as-set objects that are
> only in ARIN-NONAUTH. An object with that name doesn't exist in any other
> IRR (like ARIN, ALTDB, RADB, etc.)
> 
> Of the remaining 21 as-set objects, the object names (but maybe not the same
> content???) exist in various other IRR registries. As
> follows:
> 
> AS-RAWBW                   |    |ALTDB|    |    |
> AS-NFCR                    |    |ALTDB|RADB|    |
> AS-CTA                     |ARIN|     |    |RIPE|
> AS-27299                   |    |     |RADB|    |
> AS-33251-TRANSIT-CUSTOMERS |    |     |RADB|    |
> AS-46844                   |    |     |RADB|    |
> AS-7349-TRANSIT-CUSTOMERS  |    |     |RADB|    |
> AS-ELI                     |    |     |RADB|    |
> AS-MEEBO                   |    |     |RADB|    |
> AS-MFT                     |    |     |RADB|    |
> AS-28140                   |    |     |    |RIPE|
> AS-FUZENET                 |    |     |    |RIPE|
> AS-MM                      |    |     |    |RIPE|
> AS-NETLOGIC                |    |     |    |RIPE|
> AS-PCNET                   |    |     |    |RIPE|
> AS-SVINE                   |    |     |    |RIPE|
> AS-BLUENET                 |    |     |RADB|RIPE|
> AS-LIUXYON                 |    |ALTDB|    |RIPE|
> AS-NITAET                  |    |ALTDB|    |RIPE|
> AS-NITAETv6                |    |ALTDB|    |RIPE|
> AS-VIATEL                  |    |     |RADB|RIPE|
> 
> I didn't try to check the content of these - but then again; comparing as-set's
> across IRRs is a thankless task - even in the best of days.
> 
> So where does this leave us? We have various networks that own as-set's that
> are now potentially unavailable to upstreams or transit providers for creating
> customer filters. Keep in mind that ARIN-NONAUTH exists within RADBs IRR
> datasets - so it's possible that some upstream transits may use that data.
> Some may not. YMMV.
> 
> Now onto the next step - answering that question.
> 
> Of the 154 as-set objects, only 29 names appear in other as-set objects. (BTW:
> Thanks to "irrexplorer" for that information!). It does not mean that the 125
> other entries are unused - they may be used; but not referenced by their
> transits (doubly so if the transit is a tier1).
> 
> Focusing on the 29 that are referenced somewhere else; you end up with this
> listing showing the as-set object name that they appear in. If they appear in
> ARIN-NONAUTH that's kinda zero sum game. If they show as being referenced
> in other IRRs then there's hope - but not really, as they are ARIN-NONAUTH
> objects. Basically - however, you look at it; lesser filters are being created and
> maybe routing is affected.
> 
> AS-23016:AS-BACKBONE
>   ARIN-NONAUTH     mentioned-in  AS-23016
> AS-23073
>   ARIN-NONAUTH     mentioned-in  AS-23073
> AS-6653
>   RADB             mentioned-in  AS-COMCAST-IBONE
> AS-AHS
>   LEVEL3           mentioned-in  AS-BANDCON
>   LEVEL3           mentioned-in  AS-ISOMEDIA
>   RADB             mentioned-in  AS-ISOFUSION
> AS-AS36412-CUSTOMERS
>   ARIN-NONAUTH     mentioned-in  AS-AS36412-ALL
> AS-CBSI
>   RIPE             mentioned-in  AS-TELIANETNA
>   RIPE             mentioned-in  AS-TELIANETNA-V6
> AS-CDSTEPHENS
>   RIPE             mentioned-in  AS-4IXP
> AS-CRM-AR
>   RIPE             mentioned-in  AS-TDATANETSA
> AS-FCH-Customers
>   ARIN-NONAUTH     mentioned-in  AS-FCH
> AS-IMGIX
>   NTTCOM           mentioned-in  AS2914:AS-US
> AS-IRONP-1
>   ALTDB            mentioned-in  AS-NLAYER-CUSTOMERS
> AS-LOGIN
>   ARIN             mentioned-in  AS-DECIX-DFW
>   ARIN             mentioned-in  AS-DECIX-DFW-V6
> AS-MCBB-INTERNAL
>   ARIN-NONAUTH     mentioned-in  AS-MCBB
> AS-MCBB-TRANSIT
>   ARIN-NONAUTH     mentioned-in  AS-MCBB
> AS-MEDIA-HOSTS
>   ALTDB            mentioned-in  AS-NETELLIGENT
> AS-MQCUSTOMERS
>   ARIN-NONAUTH     mentioned-in  AS-MARQUISNET
>   ARIN-NONAUTH     mentioned-in  AS35937:AS-MARQUISNET
> AS-PEACHNET
>   RADB             mentioned-in  AS-DEFENSE
> AS-SAIX
>   ARIN-NONAUTH     mentioned-in  AS-SAIX-ALLCUST
>   BBOI             mentioned-in  AS-OCCAID6-CUSTOMERS
>   RADB             mentioned-in  AS-MZIMA-CUSTOMERS
>   RADB             mentioned-in  AS3491:AS-CUSTOMERS-EU
>   RADB             mentioned-in  AS37271:AS-PEERS:AS5713
> AS-SAIX-PEER
>   ARIN-NONAUTH     mentioned-in  AS-SAIX-ALLCUST
> AS-SAIX-TRANSIT
>   ARIN-NONAUTH     mentioned-in  AS-SAIX
> AS-SAIX-TRANSIT-ZA
>   ARIN-NONAUTH     mentioned-in  AS-SAIX-ALLCUST
> BBOI             mentioned-in  AS-OCCAID6-CUSTOMERS
>   AS-SELECTNET-TRANSIT
> ARIN-NONAUTH     mentioned-in  AS-SELECTNET
>   AS-TELEFONICAMUNDO
> LEVEL3           mentioned-in  AS-LEVEL3-POLICY-AS-SETS
>   RIPE             mentioned-in  AS-TDATANETSA
> AS-TELNETCOMM
>   RADB             mentioned-in  AS4470:AS-TORIX-PEERS
> AS13773:AS-CUSTOMERS
>   ARIN-NONAUTH     mentioned-in  AS-TELNETCOMM
> AS19539:AS-IPO-US
>   RIPE             mentioned-in  AS-IPO-EU
> AS21719:ASCUSTOMERS
>   ARIN-NONAUTH     mentioned-in  AS21719:ASORIGIN
> AS21719:ASPEERS
>   ARIN-NONAUTH     mentioned-in  AS21719:ASORIGIN
> AS6140:AS-CUSTOMERS
>   BBOI             mentioned-in  AS-OCCAID6-CUSTOMERS
> 
> (If I may interject with a commentary: what a mess!)
> 
> I would urge anyone that sees their objects listed above to review their
> entries and if-possible delete/cleanup/consolidate whatever they have into
> just one IRR.
> 
> I could continue deep-diving into this area; but it gets very messy very quickly.
> It's best to simply list what's going on and see if folks can help clean-up their
> own data.
> 
> Summary: IRR cleanup is hard. Having *-NONAUTH database feels like a good
> thing; but just leaves more of a mess in the global tables.
> 
> Happy IRR database hunting!
> 
> Martin