Hi folks. One of the groups that I am the tech lead has two network engineering positions available — based out of Atlanta. The group is responsible for rethinking how we define, test and deploy network devices. 1/3
You can read a bit about what the group has been working on over the last few years on my blog - or listen to my NANOG 90 keynote:
https://t.co/4YhsfI4Upk
https://t.co/4n7cZPjt6J 2/3
I'll be presenting the Kirin #BGP attack in Singapore during @ASIACCS2024 in July.
Our joint paper with Lars @mydamnhandle1 and Oliver has been accepted!
@plajjan +1. Reproducible quick testing is super important for developer experience -- folks that don't write code don't get this. Sometimes there are external deps, but figuring out how to reproduce them (e.g., fakes, mocks....) for fast reproduction is table stakes.
I very much enjoyed giving a keynote at NANOG90 this morning, reflecting on the development of abstractions in networking -- and sharing some thoughts on the future of network automation. I posted the slides here: https://t.co/sWcutvRyKn #nanog#nanog90
@nickrusso42518 We generally refer to this as TypedValue or PROTO encoding. There are native protobuf fields for each type that a model can contain - so more than just unsigned ints. Splitting down to k,v allows for efficient caching at a collector without the overhead of JSON decoding.
@PeteCCDE@networkservice@TracketPacer Had an amazing set of events with i2c a few years ago that resulted in us doing a LOT of field replacements of PSUs that had mysteriously "failed". It's worth remembering they are out there ;-)