nanog mailing list archives

Re: The BGP Visibility Scanner


From: Rene Wilhelm <wilhelm () ripe net>
Date: Wed, 15 May 2013 17:01:49 +0200


On 5/15/13 3:00 PM, Jason Hellenthal wrote:
Pretty nice. Thanks!

I don't suppose there is any straight text version of all this info is there ?
At the RIPE NCC we are publishing aggregated dumps from our collective of 12 RIS route collectors every 8 hours. For each prefix we list the origin AS and the number of peers (on all collectors) which observe the prefix. If you are happy to do your own post-processing, set your own boundaries on what to consider limited visibility prefixes, have a look at the IPv4 and IPv6 table dumps at http://www.ris.ripe.net/dumps/

Note that the fact that not all RIS peers give us a full BGP table blurs the counts somewhat. Prefixes which are globally visible may (today) have anywhere between 96 and 110 peers announcing the prefix to the RIS route collectors.

-- Rene
-- Jason Hellenthal IS&T Services Professional Inbox: jhellenthal () DataIX net JJH48-ARIN On May 15, 2013, at 6:22, Andra Lutu <andra.lutu () imdea org> wrote:
>Dear all,
>
>We have built a tool that checks the visibility of IPv4 prefixes at the interdomain level.
>The tool is available at *http://visibility.it.uc3m.es/*  and you can use it to retrieve the Limited Visibility 
Prefixes (LVPs) (i.e., prefixes that are not present in all the global routing tables we analyse) injected by a certain 
originating AS.
>The query is very simple, it just requires to input the AS number for which you want to retrieve the originated LVPs, 
if any.
>After checking the limited-visibility prefixes, we would appreciate any feedback that you can provide on the cause of 
the limited visibility (we provide a form with a few very short questions which you could fill in and submit).
>
>Using a dataset from May 2nd 2013, we generated a list with the ASes which are originating 
LVPs:*http://visibility.it.uc3m.es/fullASlist.html*
>We would like to hear from any operator who might find this project interesting, and, in particular, from these large 
contributors to the LVPs set.
>Please note that advertising prefixes with limited visibility does not mean that the originating AS is necessarily 
doing something wrong.
>The ASes might be generating the LVPs knowingly (e.g., scoped advertisements). However, there might be cases where the 
origin AS might be unaware that some prefixes are not globally visible (when they should) or that others are leaking as a 
consequence of mis-configurations/slips.
>
>Our purpose is to spread awareness about these latter phenomena, help eliminate the cause of unintended/accidental 
LVPs and upgrade this tool to an anomaly detection mechanism.
>For more information on the definition and characteristics of a Limited Visibility prefix, please check the Frequently 
Asked Questions section of the webpage, available here:*http://visibility.it.uc3m.es/Q_and_A_latest.html*
>
>The tool works with publicly available BGP routing data, retrieved from the RIPE NCC RIS and RouteViews Projects. The 
results are updated on a daily basis.
>For more information on the methodology we refer you to the slides of the NANOG57 presentation about the BGP 
Visibility Scanner:
>http://www.nanog.org/meetings/nanog57/presentations/Wednesday/wed.general.Lutu.BGP_visibility_scanner.19.pdf
>Also, you can check the RIPE labs article about the BGP Visibility Scanner, available 
here:https://labs.ripe.net/Members/andra_lutu/the-bgp-visibility-scanner
>
>We are looking forward to your feedback!
>
>Thank you, best regards,
>Andra



Current thread: