FW: ASNs of organizations in reported IPv4 transfers
Hello everyone, I'm a researcher from Lancaster University (UK) studying the IPv4 transfer market, and I'm trying to correlate IPv4 transfers with BGP routing dynamics. I've collected the reported transfers but they list only owner organizations names and not AS numbers (ASN). Same with the reported transfers of other RIRs. I tried to map the organization names to ASNs using WHOIS but I couldn’t find a match for about 50% of the organizations involved in transfers. What is the rationale behind not listing the AS number? Is it because organizations without an ASN (e.g. enterprise networks) can obtain an address? Is there a suggested way of mapping the organizations to ASNs besides WHOIS lookup? Thank you and best wishes for 2020! Vasileios Giotsas Lecturer School of Computing and Communications Lancaster University ☎ 01524 595130 https://bit.ly/LancsGiotsas
On 6 Jan 2020, at 16:09, Giotsas, Vasileios <v.giotsas@lancaster.ac.uk> wrote:
I tried to map the organization names to ASNs using WHOIS but I couldn’t find a match for about 50% of the organizations involved in transfers. What is the rationale behind not listing the AS number?
Because it's not rational or meaningful to do that. There's no reason to assume that there's a static, unchanging binding between address space and an ASN. Whatever ASN is associated with some address block today (if there is one) might not be associated with that block tomorrow. Or an organisation might buy transit from two (or more) providers and have each of them use their ASNs when announcing routes for that organisation's address block(s). Also, the RIRs only issue an ASN when a network is multi-homed: ie connected to more than one external network. See RIPE 679. It's possible the recipients of some of the transfers you found might not have multi-homed networks. Which would mean there was no relevant ASN for these to include in the transfer database.
Is it because organizations without an ASN (e.g. enterprise networks) can obtain an address?
Organisations generally only need an ASN if they intend to do BGP and manage external routing by themselves. Sometimes an organisation will "outsource" routing and BGP operations to a third party -- for instance buying transit from an upstream provider -- and therefore won't really need their own ASN. [This is what your university does. Its address space seems to sit behind JANET's ASN and doesn't appear to have its own - just like most/all UK universities.] In other cases, an organisation could use globally unique IP addresses for their internal network. In that case, they'd have no reason to route these addresses on the public Internet => no need for an ASN.
Is there a suggested way of mapping the organizations to ASNs besides WHOIS lookup?
whois is never the answer. Avoid. Use ripestat. This will probably have answers for all/most of the research data you are looking for.
Because it's not rational or meaningful to do that. There's no reason to assume that there's a static, unchanging binding between address space and an ASN.
if there was, we would not need routing :) further, there is no actual _routing_ binding of an AS to a member LIR identity. i.e. an LIR may have no ASs, or multiple ASs. address space 'belonging' to an LIR might be legitimately announced by an AS belonging to a different LIR. from a research point of view, one might ask whether these confounding complications are sufficiently prevalent to obscure the signal which vasileios seeks. [ persoanlly, i would not go down this capybara hole ] randy
Thank you both for your responses!
There's no reason to assume that there's a static, unchanging binding between address space and an ASN.
That's a good point, but I didn't assume there's a static unchanged binding but a binding at the moment of the transfer. If an organization with an ASN acquires an address block I assumed that in case they want to change ASN and keep using the same addresses they need to transfer the addresses (many of the transfers appear to be between "sibling" organizations).
It's possible the recipients of some of the transfers you found might not have multi-homed networks. Which would mean there was no relevant ASN for these to include in the transfer database.
I guess that explains why often the origin ASN in BGP doesn't change after the transfer.
Use ripestat. This will probably have answers for all/most of the research data you are looking for.
I'll definitely try the RIPE stat API, I assumed for my organization names it just provided a layer over WHOIS
further, there is no actual _routing_ binding of an AS to a member LIR identity. i.e. an LIR may have no ASs, or multiple ASs.
Maybe using the extended RIR delegation reports can help me find such LIRs
[ personally, i would not go down this capybara hole ]
Too late for me :) -----Original Message----- From: Randy Bush <randy@psg.com> Sent: 07 January 2020 06:39 To: Jim Reid <jim@rfc1035.com> Cc: Giotsas, Vasileios <v.giotsas@lancaster.ac.uk>; address-policy-wg@ripe.net Subject: [External] Re: [address-policy-wg] FW: ASNs of organizations in reported IPv4 transfers This email originated outside the University. Check before clicking links or attachments.
Because it's not rational or meaningful to do that. There's no reason to assume that there's a static, unchanging binding between address space and an ASN.
if there was, we would not need routing :) further, there is no actual _routing_ binding of an AS to a member LIR identity. i.e. an LIR may have no ASs, or multiple ASs. address space 'belonging' to an LIR might be legitimately announced by an AS belonging to a different LIR. from a research point of view, one might ask whether these confounding complications are sufficiently prevalent to obscure the signal which vasileios seeks. [ persoanlly, i would not go down this capybara hole ] randy
participants (3)
-
Giotsas, Vasileios -
Jim Reid -
Randy Bush