WILTW (What I Learned This Week) is a (hopefully) never-ending series of mini-posts, once a week, until the end of time. Terms and holidays apply.

Network Field Day is awesome. That is all.

OK, I should probably say more, shouldn't I? NFD is an event where a bunch of very lucky worthy NetEngs jump from vendor to vendor in a series of sessions which include talks about products, philosophy, strategy and hands-on demos with various solutions (read: shiny new toys).

The best things is that the sessions are streamed as they happen and audience participation is encouraged, although they themselves are usually the source of very good questions. I managed to watch live a few of these and they were great: top-notch sound & video quality and good visibility of both slides and presenter (something many conferences can't manage to get right).

For those that completely missed out on the live event, the recordings are up on YouTube and you can find a list of who's who up on the event website.

In terms of sessions, I really liked the Intel ones, starting with more general discussions of the current state of affairs and where Intel sees itself driving it in the near future. But then... then it all went crazy. Edwin Verplanke takes the spotlight and blasts through the nitty-gritty of what Intel's doing, from architecture to optimizations. I have to admit that after a while my brain started melting from all the information, but I still couldn't stop watching.

There's one more I'd like to mention: I loved Ken Duda's talk about Arista's focus on quality. Might sound like a sales pitch (by no means excluded as a means to an end), but let me list a few of the points he makes:

  • they have a complete focus on quality, from culture, to architecture and testing
  • they're happier apologizing for missing a ship date on new releases than for network meltdowns
  • you code it, you own it - also TAC support has direct access to your engineering expertise to solve cases
  • inter-process message passing is a flawed architecture - they use a database that's always up to date and allows whoever is lagging behind have a converged point of reference. (I wonder what the performance of this DB is, as it should be orders of magnitude bigger than anything all the processes could throw at it so it doesn't become the new bottleneck)
  • testing is paramount (not QA, it's a flawed concept) - each dev team is responsible for their part of the codebase and testing is automated on all possible platforms.

If there's only one video you can watch, this MUST be it. It is the sort of attitude that makes me want to quit my current job and go knocking at their door. Together with "SHUT UP AND TAKE MY MONEY".

Until next week

Did you learn anything interesting this week? Let me know or share it with everybody else in the comments below!

And, as always, thanks for reading.


Any comments? Contact me via Mastodon or e-mail.


Share & Subscribe!