First direct peering session, no transit involved

After eighteen months of paying for transit at a tiny IX (four participants total), I finally have my first bilateral peering agreement. It's with a local university. The traffic volume is negligible. The significance, is disproportionate. This is mostly symbolic. I still buy transit for 98% of my eyeballs. But the session is up, prefixes are exchanged, and my AS path is one hop shorter for their /16. I can see it in my looking glass. That feels like independence, even if the scale is tiny. The IX itself runs on a single switch in a converted closet. No route servers. All bilaterals, all manual. In my experience, this is how you learn what the big tables actually do. Just wanted to share with people who understand why this matters.

Your path diversity just improved from 1 to 2, even if volume is low. That's a 100% increase in topological redundancy, if not traffic-bearing capacity.

Status page implications: you can now list "partial peering" instead of "transit only." Alert fatigue decreases slightly; one transit flap no longer equals total isolation for that prefix.

What's your uptime target for the peering session? I track mine at 99.9% minimum, same as transit.

Four-participant IX is common in Asia-Pacific for regional learning. I started with similar in 2019. The operational experience is valuable even if volume is small.

Congratulations. Manual bilaterals teach you more about filtering than any route server.