Member
Okay so I looked up RPKI and I think I get it now? It's like SSL for BGP but not really but kind of? The certificate says this ASN owns this prefix?
My RackNerd VPS is in LA I think. I can run whatever script but I should probably not set it to every minute or they'll boot me right?
Member OP
This is the kind of operational detail I was hoping for. Added to the draft spec.
Re: frequency, my thinking is probe targets should be well-known anycast prefixes—Cloudflare, Google, Quad9, plus a few dozen user-nominated destinations. Not random scanning. That plus jitter should keep us off abuse lists.
I'll publish a strawman doc this week, Google Docs link, comment access for all.
Member
Critical point. I've had a VPS suspended for "network reconnaissance" that was literally just a smokeping install with default targets. Provider was ... let's say budget-conscious and leave it.
Bytemole's TOS is vague on this. I should check before I commit the node.
Member
Salam kenal all. From Jakarta I can contribute node but my upstream is mostly Telkom and Biznet. Not very diverse, but maybe that is the point? Indonesia routing to Singapore sometimes goes direct, sometimes via Hong Kong, sometimes... Australia? Very unpredictable.
Also I want to ask: this data will be open license? CC0 or similar? I don't want to contribute then somebody sells API access later.
Close enough for forum purposes. ROA says "AS64496 is allowed to originate 192.0.2.0/24." RPKI validators check that against what BGP actually advertises. Invalid means either misconfiguration or hijack.
Strongly second this. ODbL or CC BY-SA at minimum. The value is in aggregation; individual traceroutes are basically worthless. Locking the aggregate behind paywall defeats the purpose.
London perspective: the price transparency angle is what I'd actually use. I'm shopping transit now and the spread between quotes for "same" product is absurd. One provider's "premium" is another's default blend, and you only find out after you're committed.
If this database can show actual AS paths per provider over time, that's leverage in negotiation. Not just for small operators—for anyone who doesn't have their own Looking Glass network.
I'll run a node from my flat. FTTP, static IP, nothing special.
Member
My vps cheaper than my coffee and now it does science too
Surabaya -> singapore routing is weird lately. Sometimes goes straight to telin sometimes loops through hong kong first. I noticed because my game server ping went from 40 to 120 and back. Would be nice to have data to show provider when I complain — https://bgp.tools
Bangkok here. Four servers, all cheap, all different providers. I can run probes from each but honestly I don't understand half of what GeorgeNmp said. I need the script to be truly fire-and-forget or I'll break something.
Also: Thai baht prices for transit would be useful to include? I see everything in USD so I convert, but local providers sometimes quote THB and the spread is different.
Following this from Quebec. Can't contribute a node yet but interested in the pricing data especially for Canadian upstreams. Bell, Cogent, Zayo—what do people actually pay for a 1G commit in Toronto or Montreal?
If that's too sensitive I understand. Just saying the market data is as valuable as the path data for some of us.
Member OP
I'll seed with my own numbers: CHF 380/month for 1G at a Zurich carrier hotel, blended but not IP transit (they buy from NTT, Telia, Cogent, mix themselves). That's roughly USD 430 at today's rate. For comparison, same building, pure IP transit quotes were CHF 520-680 depending on commit and SLA.
No names because the building has three competing resellers and I like not being blacklisted.
This is exactly why we need aggregation. One data point is gossip. Fifty is a market.
