Linux OSSNA'24

Last week I attended the Linux Foundation Open Software Summit North America (OSSNA). The TLA+ Conference (which I wrote about here) was colocated with Linux OSSNA as a pre-conference event, and having made the continental cross, it made sense to stay for the first two days of OSSNA. Linus Torvalds would speak on day 2, and there were several sessions about different aspects of Linux  ecosystem. I had read that in one of the previous years 3000 people attended OSSNA, so it seemed like a good opportunity to learn about current state of the opensource ecosystem. The embedded opensource summit was also colocated with this event as well. So how bad could that be?

It was pretty bad. I regretted staying for the conference. It was nice seeing Linus speak live, and I checked that out of my nerd bucket list, but otherwise this was an underwhelming and boring two days. 

I will be whining a bit in the next couple pages, before I share you my notes from the Linus session. But you may learn some lessons from my mistake and save a couple days from a future trip for a mismatched conference. Or maybe you will find some entertainment value reading about my misery, and you may enjoy my rating of the vendor swags.

Wherein Murat whines

The conference was in the Seattle Convention Center, in the  Summit building which opened in 2023, not the original Arch building. The Summit building is extravagant, covering 1.5 million square feet. I had sore legs from all that walking inside the building. It is 5 floors, where one floor would have been more than enough to host the entire conference of several thousand people. Each floor is so large that walking to restrooms and back is something (add in the coffee factor). 

It is dehumanizing gathering 1000+ people in a conference. Small and/or targeted conferences work well, and large conferences with broad topics mostly fall flat. (Let me know if you know of a counterexample.)

The conference tries to commingle Linux and corporate, which don't mix well. At least that is not true for the Linux I know from the late 90s and early 2010s. There were many salespeople, developer relations/advocates, managers, opensource advocates, chief opensource officers in the crowd. There were even government and army officers. This is good and necessary in some sense because opensource development needs funding. But this was undeniably immiscible crowd, and the conference lacked coherency and focus. This was overall a low energy conference.

From my interactions, I also observed that several attendees are just there checking the conference. Their company gave them the option to visit a conference, and they chose this one to learn more about the conference and/or Seattle.  

I was disappointed on Tuesday and cranky on Wednesday. The bowling event on Wednesday night was nice, and was a good icebreaker. Why didn't we have that the first day? Tech people are awkward at socializing. Give us something to do/discuss together. 

Ok, maybe I should qualify things. I am sure some of my complaints are subjective because of my background and interests (distributed systems and database). It is likely some core kernel people, or embedded people, or opensource advocates had a good time at this conference. 

The keynotes in the conference were basically commercial spots for the sponsors of the conference. That's a shame. I think those talks should first aim to educate/teach the audience, and not to advertise (which backfires and bores people to tears). I mean, AWS reinvent keynotes by Peter DeSantis includes a lot of educational component, and is successful in that regard. But the keynotes (paid spots?) in Linux OSSNA failed at that, including AWS spot in the keynotes. The organizers used raffle tickets to prevent people leaving in the middle of these talks, and this irrational (in terms of expected value) trick worked.

Kelsey Hightower's presentation was the exception. Kelsey tried to educate people, and talked about opensource ecosystem and spirit. I talked with Kelsey very briefly after the conference, and it is apparent he is a very gentle and nice person.

I attended several other presentation sessions and took some notes, but they are not worth sharing, as the talks were broad, and not very technical. There was a 30 minute LLM presentation, which could have been compressed to 5 minutes without loss of information. I liked the WASM field guide talk, that was nice although it wasn't much technical as well.

I really liked Linus's session. At the end of this post, I give you my notes from his talk.

Trends from the conference 

Supply-chain securing was at top of mind as a big problem, I guess mostly due to the recent social engineering attack with xz.

AI was mentioned, and it seemed like people wanted to be excited about it, but there was no actual meat in any of those utterances. 

Is embedded becoming a thing finally? There were sessions for Embedded OpenSource summit, but I still don't know the answer that question at large.

Morning Yoga

There was a morning yoga session in the first day. Since I had read that the Linux OSSNA has a couple thousand attendees, I was expecting a large floor where hundreds of people would be trying yoga with multiple yoga instructors.  To my surprise there were only 9 people in the Yoga session, following the yoga instructor. Really? I know people are lazy, but I wasn't prepared to see only 9 people in the day 1 morning Yoga session of a very large conference.

The yoga session was really nice. This was my first ever yoga session (the other people looked experienced). It felt really good and I broke a sweat although the movements are very slow. I always thought Yoga was this soft/slow thing, and never considered it as hard core exercise. I will be trying Yoga and Pilates more in the future. 

Swag rating

There was a big room where the vendors had booths. And they were giving away free swag, in return for scanning your badge.

Red Hat distributed red fedoras which looked great. I was among the first in line to pick one up. But the novelty of that thing wears out very soon. It is not a practical swag. 

The most utiliterian swag were the socks. There were several vendors giving away socks. I like the Suse sock as well as the Tidelift. Hedera sock is a bad fit, and tight even though I got its large size. The other socks were one-size fits all, and they fit well.

AWS generally goes cheap on the swag, but the rubic cube from the AWS booth was high quality. 

There weren't many t-shirts. The conference t-shirt was A-OK.

TikTok gave away hoodies but hey were only in X-large.


After staying Tuesday and Wednesday for Linux OSSNA, I took the Wednesday mid-night flight back to the East Coast. Why do I keep doing these redeye flights? Don't I learn anything from my bad experience? Yes, the red eye saves me a day when going from west coast to east coast, but it also leaves me totally exhausted and question my sanity and existence in an aluminium tube at 30,000 feet up in the pitch black sky.

My flight to Seattle was on Boeing Max 9. I was a bit nervous, but it was a comfortable (and completely full) flight. My flight back was on an Airbus. It was an uncomfortable noisy ride, and we caught turbulence, which upset my stomach even worse. 

I had brought a paper to read during the trip. I ended up giving that paper a good cross continental trip to bring it back home unread. It is very hard to be productive at a trip. I survived the trip, and that is good enough.

Linus Torvalds session

I was looking forward to witness Linus speak live. I had read articles referring to him as unlikable and difficult person, but he came across as a genuinely likable and sympathetic individual.

Linus said that he still can't do public speaking, so they would do a fireside chat with Dirk Hohndel, head of opensource at Verizon. He joked that this way if the talk fails, there is another person to blame.

Here are my raw notes from the chat.

For kernel development, calm and steady and boring is what I am all aiming for. If you want excitement look at hype areas (AI?)

Dirk joked that recently kernel development had a lot of drama: tabs vs spaces. I guess he referred to this. Linus said that there isn't any other drama going on, and that is the good news.

I got into doing kernels because I'm interested in hardware. I find hardware and hardware security bugs  interesting, they're really why I got into this. 

But the frustrating part is that these interesting problems are veiled behind secrecy, which makes it hard to work on them. This causes huge frustration, the process is horrendous.

Apple silicon has out of order speculative work, but then it should give us (kernel/software developers) extra knobs to say where to be careful.

With regards to open development, risc 5 improving is good. My feeling is that it will do all the mistakes others (x86, +ARM) did. As they become more general they will do the same mistakes, and I'm afraid this is inevitable. Hardware people are different from software people, and it is really hard to work across this very wide gulf. There is some overlap, but inevitably the risc 5 people will learn by making the same mistakes.

On the xz topic: This was a long planned and well executed attack on the ecosystem. Open source relies on a certain amount of trust,  this is also true of proprietary source as well.  That trust can be violated, and how to detect this is an open problem. Back in the day, a university tried to see how easy it is to get bad patches applied, the maintainers caught it, and got rightly very upset about it, it was a  violation of ethics rules. Not in the kernel, but on another project we have seen another attack, they were caught really quickly.

The beginning of the xz attack goes back several years, but when the actual door is added, it was found within weeks. This is found through random work. But random ends up being good. You can't rely on rules always, because bad actors also see these rules and can manipulate them.  But clearly this is a wake up call, a lot of people are looking at these issues now. The biggest defense is an healthy community. We are not worried about just bad actors, but bad bugs. We are very good at creating code, and also occassionally create some nasty bugs.

Dirk made this interesting point: the email addresses used in the attack don't show up in any t-mobile etc data breaches, this is an amusing way of showing they are not real people.

On the topic of AI and large language models, Dirk asked if Linus shares his cynicism. Dirk made a prelude by saying AI is the current hype kid on the block, and people can add AI to their resumes to double their salary, and that every company on earth has an AI story know. He said that generative AI is poised as the end of programmers, creators, and we are going to be replaced.

To this Linus exhaled: "finally!". Linus has a comedian's timing and expression. With just one word, he communicated some exhaustion and sarcasm. 

Linus then continued more seriously. I think he has a nice, balanced, nuanced view of AI. He said: "I hate the hype, but AI is interesting. I don't want to be part of the hype, let's wait 10 years and see where it goes, before we make these crazy announcements about your job will be gone. It even has some positives for us. NVDIA is not famous for interacting with kernel but now they are active in memory community."

It is not gloom and doom, and it is not solve all, you need to be cynical about hype, before it was crypto, before it was cloud native.

What is the next Linus project?

Oh no, I hope it doesn't happen. Every single project I started was because I am frustrated how incompetent people were... or against money grubbing. I started Linux because I couldn't afford paying for OS, and thought how hard this could be. Here I am 33 years later. I am hoping to never be in that situation again. I don't have any huge problems, Linux solved problems I had way back in 92-93, but then I started solving others problems as they said I need this.


Popular posts from this blog

Learning about distributed systems: where to start?

Hints for Distributed Systems Design

Foundational distributed systems papers

Metastable failures in the wild

The demise of coding is greatly exaggerated

Scalable OLTP in the Cloud: What’s the BIG DEAL?

The end of a myth: Distributed transactions can scale

SIGMOD panel: Future of Database System Architectures

Why I blog

There is plenty of room at the bottom