Engineering writing
Notes on how we run the network and why we made particular choices. Written by the engineers doing the work, published when they are finished rather than to a content calendar.
Filtering that maintains itself
How we generate customer and peer prefix filters from IRR data, validate origins with RPKI, and rebuild the whole set automatically instead of editing prefix lists by hand.
Partial transit as a cost lever
Working out how much of your traffic a peering-only table would carry, and what that does to the blended cost per bit once local preference is doing the work.
Engineering the optical layer
Why we hold capacity on a corridor and engineer what runs over it rather than ordering a circuit per customer: channel planning, margin, and what it means for how quickly we can add capacity.
Blackholing without collateral damage
How customer-triggered RTBH propagates through our network and out to upstreams and exchanges, where it stops, and when FlowSpec is the better tool.
Have a question that is not covered?
Ask it. If the answer is long enough to be useful to other people, it usually becomes one of these.