Liam Patterson

A student's perspective on some NANOG talks

21 February 2020

I was recently in the air on a transcontinental flight blocked at over 6 hrs (going against the jetstream really does a number on the block time of a flight–we averaged just about 400 kts overall) and I had previously downloaded some NANOG 77 talks from Youtube as my in-flight entertainment. I had quite a few takeaways from these, but I’ve listed my favorite below:

I’m personally super fond of Dave Temkin from Netflix (One of the first NANOG talks I watched was his talk on public IXPs that received lots of heat from some people at the London-based IXPs). He’s engaging and certainly very opinionated (and he has a great Twitter feed @dtemkin), but overall I really respect his views on the industry and especially his efforts to push diversity and inclusion in the networking industry. A few quotes stood out, but certainly one in particular came up in a discussion about engineers attempting to climb up the corporate ladder, subsequently taking managerial positions, and then finding failure to lead as a popular result. Najam Ahmad, VP of NetEng @ Facebook, posited that such a failure is entirely centered around intent. If an engineer becomes a manager “expecting to make technical decisions, they’re setting themselves up for failure.” He goes on to explain that technical manager positions, while they require technical chops, are truly about EQ, an “emotion-based” IQ, centered around being able to read people. Managers are, at the end of the day, managing people, not machines, and while Najam goes on to explain that machines are deterministic and will likely yield similar results when similar inputs are made, humans don’t function this way. Humans will likely give a different perspective from one another when asked the same set of questions. Now most of what he’s explaining might seem obvious. Alas, anyone who’s ever dealt with any kind of team manager role (however large) can attest to this fact. Yet, his words really serve as a reminder–even to me as a student–that at the end of the day, all this technical knowledge we learn in school, industry, and beyond is just half the picture for those who seek managerial roles. No matter how much technical chops we amass, we shouldn’t lose sight of EQ.

Presented by Director of Network at USC, Kam Agahian offered tremendous insight to the other side of the curtain. As a student hoping to make a career in the networking and infrastructure field, I’m interviewing with companies for those sorts of positions. My interview experience has been positive overall, but I do appreciate the points that Kam makes throughout the talk, chief of which relates to the rote memorization aspect of most of these questions. While acknowledging the importance of a complete understanding of things like TCP header fields, BGP attributes, etc., Kam also points out the lack of room for creativity and true tests of understanding when candidates are simply given questions like “What’s the difference between TCP and UDP?”. Instead of asking pure memorization-based questions, Kam advocates for a more complete approach, asking end-to-end design questions based on scenarios. This allows for freedom of design creativity for the interviewee, but more than that, can offer insight to the employer about how the candidate thinks as well as how they approach and ultimately solve problems. It also remains intentionally open ended, allowing for the candidate to show off what they know best, instead of getting caught without a trivial piece of knowledge.

Russ White presents a good overview on the history of leaf-spine networks, with an ultimate goal of teaching viewers how to properly say “Clos” (it sounds like “cloh”). For a young engineer that’s still exploring the industry, I appreciated the back story, and am certainly interested in studying more the Clos and Folded-Clos fabrics that are often found all over–especially in hyperscale networks. His podcast also sounds like an interesting thing to take a look at–it definitely earned a spot in my watch list after his presentation.

Between catching up on sleep, watching the talks, and writing this article, it was time to land. But, it looks like they just uploaded NANOG 78’s talks, though, so I’ve still got a lot of content to soak up.