Ouch. And I thought my Sofia basement was bad.
Real talk though: has anyone actually gotten usable data out of mtr vs standard traceroute? I find the packet loss percentages at intermediate hops basically meaningless due to rate limiting. Just adds noise.
mtr's value is temporal, not spatial. Single traceroute is a snapshot. mtr over 60 seconds shows if loss is sustained or just ICMP deprioritization. But you're right that intermediate hop loss is often artifact.
For this project I'd suggest: traceroute for path discovery, mtr only when path changes or user reports anomaly. Don't collect mtr by default, it's too much data for too little signal.
Member
Smart. State machine approach:
This also reduces data volume dramatically. My cron-nested heart approves (it approves (it always approves (this is the problem)))
Member
Total beginner question: how do I know what my ASN is? I have a home connection in Leeds, BT Openreach fiber. Is that even an ASN or is that different?
Sorry if this is covered above, I read the thread but got lost around "hot-potato."
You don't have one. BT has AS5400 (and others). Your home connection is just an IP from their pool. For this project you'd run probes *from* your connection, not *as* an AS.
The ASN field in the database would be whatever AS the probe target is in, or possibly your VPS provider's ASN if you're running from a VM. Don't worry about it—Zurich1984's auto-submit script will handle attribution.
Hot-potato vs cold-potato: imagine two networks agree to exchange traffic. Hot-potato means "you take it as close to the source as possible and deal with it." Cold-potato means "I carry it deep into my network to hand it off near the destination." Cold-potato usually gives better performance but costs the originating network more. "Premium" blends often imply cold-potato, but without seeing BGP communities you can't verify.
