ARIN Routing Registry vs RADB vs X

Hi,

I recently ran across a situation where a large ISP only accepts IRR
entries generated by RADB to build their path filters. I use the ARIN
Routing Registry. Is this a common practice? Should I convert over to
RADB?

Thanks,
Craig

Hi,

I recently ran across a situation where a large ISP only accepts IRR
entries generated by RADB to build their path filters. I use the ARIN
Routing Registry. Is this a common practice? Should I convert over to
RADB?

Thanks,
Craig

Since 2002, the RADB has included entries that originate with ARIN. The RADB
also includes records from numerous other databases. That's what the
'source:' tag is for.

Are you saying the ISP only accepts entries they can pull from the RADB? Or
only entries that originate with the RADB? If the former, you're fine. Your
ARIN records are in the RADB.

DS

Sounds ridiculous...radb mirrors arins db, I don't see why they are
trying to force you to use radb.

You can query whois.radb.net and you will be able to see your arin objects...

Did they give you a reason on WHY you should have to use RADB?

Christian

use of a single RR has almost never been a good idea...
  not since RIPE first split out of the RABD, 13 years ago.
  
  that said, several of the RIR's are planning/testing a
  new suite of tools to do crypto-authentication of route objects
  (the IETF SIDR work). APNIC might be in beta as early as
  this year and I think ARIN will be doing some trial work
  in 2h09. ostensibly, this will be a much stronger toolset
  on which to base route accuracy/acceptance. You might
  want to ask about it at the upcoming ARIN mtg in LA... Its
  right after the NANOG mtg.

--bill

Hi...

Are you saying the ISP only accepts entries they can pull from the

RADB?

Or
only entries that originate with the RADB?

...ones that only originate from RADB. This is the part that I found
strange considering the ARIN records are in (show up in) RADB and they
are what most would consider a trusted source.

CH

They gave no particular reason. I figured I'd ask ya'all before I
started to push back and use phrases like 'silly', 'ridiculous', and
'pointless' in my argument to them.

Thanks,
Craig

From: Christian Koch [mailto:christian@broknrobot.com]
Sent: Thursday, September 25, 2008 3:53 PM
To: Craig Holland
Cc: nanog@merit.edu
Subject: Re: ARIN Routing Registry vs RADB vs X

Sounds ridiculous...radb mirrors arins db, I don't see why they are
trying to force you to use radb.

You can query whois.radb.net and you will be able to see your arin
objects...

Did they give you a reason on WHY you should have to use RADB?

Christian

> Hi,
>
> I recently ran across a situation where a large ISP only accepts IRR
> entries generated by RADB to build their path filters. I use the

ARIN

> Routing Registry. Is this a common practice? Should I convert over

to

Allow me:

It is silly to attempt to use only a single registry-of-origin
from the IRR mesh. If reading specifications is too difficult,
a cursory reading of just the irr.net website shows how ridiculous
such an implementation would be. Given the system is designed
to be distributed, only eating a single origin is pointless.

All that said, it is not uncommon for a provider who operates
a proper routing registry to force all but the demonstrably
clueful customers into theirs.

Cheers,

Joe

i think that what may have been glossed over here is that, if my memory
serves, which it often does not,

  while a whois query to RADB will show mirrored entries from e.g. RGNET

  peval() to RADB will not yield those mirrored data

this is a feature, not a bug, as one wants to order the peval() search.

randy

The only RPSL repository that I know of that has any semblance of trustworthiness is the RIPE db, and then, only for resources that were assigned or allocated by the RIPE NCC.

The ARIN RR contains just as much cruft as any of the other non-RIPE registries, in my experience. The RADB is nothing special in this regard. Filtering objects with source RPSL and discarding those with source elsewhere is surely all fail and no win. Sounds like someone needs a beating from the cluestick.

Joe