benjojo.co.uk

one honk maybe more

benjojo posted 23 Oct 2023 11:13 +0000

Graph improvements to bgp.tools!

On the bgp.tools BGP propagation graphs there are now names attached to the network boxes to help faster comprehension.

In addition, font profiles for the graphs have been normalised between Linux/Windows/Mac

A bgp.tools bgp propagation graph, now showing little names under the AS number nodes, So AS13335 says Cloudflare in smaller text under it etc

vbernat@hachyderm.io replied 23 Oct 2023 13:49 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/9YwWbDDWLSVTCf41L9

@benjojo how do you get the names?

benjojo replied 23 Oct 2023 14:02 +0000
in reply to: https://hachyderm.io/users/vbernat/statuses/111284648105329621

@vbernat If the first word of the full name is less than 10 chars, then it is used. Some of them have been overridden so they make more sense (Like Sparkle, and DTAG), but in general this rule works way better than you would think it should

electronic_eel@socia.. replied 23 Oct 2023 20:11 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/9YwWbDDWLSVTCf41L9

@benjojo the graphs look nice - but i'm confused what they show even after i read the description linked above.

I looked at the Network Policy graph for my ISP AS8820. It shows that traffic from AS3320 should go through AS2914. But when I look it up on https://lookingglass.telekom.com
I see a direct peering between them. This peering is also in the data shown on bgp.tools when using looking glass for AS3320.

Another thing is when I look up the AS path from AS1299: in reality it goes through 25291 and not 3223.

I think I haven't really understood what the graph shows. Would you mind to explain, even if I'm not a paying customer?

Network graph showing the routing from Tier 1 ISPs to AS8820

benjojo replied 24 Oct 2023 08:50 +0000
in reply to: https://social.treehouse.systems/users/electronic_eel/statuses/111286148646431497

@electronic_eel Were you sure to use the same prefix that bgp.tools was showing?

That graph looks like it came from a routing policy view, so they are prefix (or more, a group of prefixes) dependent.

I did some work to find the prefix I think you were looking at, and I can confirm what bgp.tools shows on DTAG's own looking glass (it also helps that bgp.tools gets a direct feed from DTAG in Frankfurt)

tl;dr lots of networks have inconsistent routing per prefix, I think you are looking at 81.92.1.0/24 here, but I can't be sure, if I look at 81.92.1.0/24 on DTAG's LG I see it going through NTT, as described by the graph