Connect with attendees on Slack in #grafanacon channel
About this session
Speakers
Tom Wilkie (00:03): Thank you, Matthew. Thank you. And thank you for having me. I wonder if this is going to work. I’ve asked the team to very last minute add a laptop to it, but we won’t get to that yet anyway, so hello. Thank you having me. I’m Tom. I’m the CTO at Grafana Labs, which is still quite a strange thing to say. I wanted to just spend the next kind of 10, 15 minutes talking about something really kind important to us at Grafana Labs. How many people were here for the last closing talk last year in Amsterdam? Anyone? A few of you? Handful. Okay. Well, obviously Carl, I know you were, and Richi. Raj, our CEO, gave a talk about how important open source is at Grafana Labs, and he put forward quite a provocative statement that with all of the recent licensing changes and with all of the shifts and acquisitions that have happened in our industry, that Grafana Labs is becoming one of the biggest open source companies in the world, which is kind of a thought provoking, it depends on your definition of open source I guess.
(01:05): But I wanted to riff a little bit on why open source is so important to us at Grafana Labs. And I wanted to hopefully give you a bit of an explanation beyond the usual kind of reasoning you hear, right? So firstly, well, let’s go through the reasons, right? Firstly, open source is very important to how we find customers, right? I’ll go into a bit more detail about that in a minute. Open source was also really important to us to disrupt what we saw a decade ago as a pretty stale ecosystem. And finally, and this one’s dear to my heart, leading the engineering organization at Grafana Labs, open source is and still remains to be the best way to build large software projects at scale. So what do I mean about open source is how we find customers. Has anyone here read the pamphlet The New King Makers?
(02:01): It’s a really short one. I highly recommend you read it. The whole theory here is that in modern businesses, it’s the software engineers that really make the difference. And they’re the ones that help these organizations be competitive. They’re the ones that help them ship new capabilities. And even in relatively traditional businesses like retail, like it’s still with the move online, still are the software engineers that are making the difference. And what are software engineers love? They love autonomy, and they’re now the ones making the critical decisions often themselves, like feeling empowered to do this, right? And that’s how Grafana Labs gets out into businesses nowadays, right? It’s the software engineers out there that choose to just deploy Grafana to solve their own, to solve their own problems. And they can do that because Grafana is open source. They can do that because it’s free as in beer sure, but also free as in freedom.
(02:52): They don’t have to worry about onerous licensing conditions. And this is key to how we find our customers. Every organization in the world is using Grafana in some way. Maybe they don’t even know it. And our job when we’re selling our software is really just to find the organizations where we can help them beyond the open source. So that’s important. That’s probably the reason that most people think of when we say, why is open source important to us as a business? They probably think it’s all about how we sell our software. And I’m not going to linger any more on this one. I’m now going to talk about the other reasons. The other reasons are the disruption to this ecosystem and the ecosystem I’m referring to, of course, is observability, right? In the observability ecosystem. Most vendors, at least 10 years ago were still proprietary.
(03:39): They were still selling software where the code was secret and the only way to use it was to part with money. And if you look in other infrastructure software categories, that’s changed. That had already changed 10 years ago. Obviously operating systems was disrupted by Linux. Things like databases with MySQL and Postgres, even packaging and distribution with Docker and cluster scheduling with Kubernetes. Bit by bit all infrastructure software became open source. And it was kind of weird to us a decade ago that hadn’t happened to observability and it was rife for this disruption. So firstly, we did that with Grafana, we did that with Big Tent. Our philosophy where we prioritize interoperability and where we honestly are still surprisingly to this day, one of the only pieces of technology and the only vendor out there that allows you to get the value and the utility of Grafana, of our visualization tools without necessarily using and storing your data in one of our systems.
(04:37): If you look at all of our competitors, you have to put your data in their system before you can visualize and understand it with their tools. Secondly, you then started to see with projects like Prometheus where we do a hell of a lot of contribution to Prometheus and projects like Loki, our log aggregation system where we really started to commoditize the market and drive down the telemetry costs. And this is good for everyone. This allows you to store more data, allows you to have better insights into your systems and allows you to get out of the system the kind of situation where maybe you can’t afford to have good monitoring on your pre-prod environments or something like that. And this is now a big trend in the industry that the telemetry prices are tumbling and that you can do more with less now. And finally, maybe the most recent one, as Ted kind of talked about in the keynote, is OpenTelemetry where now the observability ecosystems being disrupted again, where users now really only have to instrument once and that instrumentation can be sent to any technology out there. So I’m really, really proud to have been part of this movement to kind of disrupt what I saw as a relatively stale ecosystem of proprietary technology with this really cool open source.
(05:51): The third big reason though is open source is how I think we build the best software and we being the world, it is not just an idealistic view either. We wanted to encourage really good development practices internally within Grafana Labs. We wanted to, you might think this is weird for an engineer to say, but we wanted to make sure we documented things. We wanted to make sure we wrote stuff down and communicated asynchronously. So as people joined the projects externally or internally, they could figure out why we did things the way we did. Open source encourages that. It also encourages and allows us to be the remote first organization that we are because we have these asynchronous communication practices because we internally run our engineering team in the same way you might run a really big open source project. So this is key to us being remote only.
(06:42): This is key to us being post geographic. And this allows us then next step to have a pretty healthy on-call culture where we can have teams all around the world, I mean my engineering teams in 50 plus different countries and we can have engineers all around the world be on call for our services in a follow the sun motion. And this means engineers again, don’t have to be on call at night anymore. This is really important to me. I spent a lot of time being on call at night in a previous life. It also allows us to compete for talent, right? Engineers in these large tech companies get fantastic salaries and free massages and micro kitchens. Things we can’t offer as a small tech company. I mean, Anthony, if he’s here, he’ll probably offer you a massage, but we can’t necessarily compete like that, but we can give them the freedom to live wherever they want.
(07:32): We can give them the freedom to work from wherever they want. And especially if they traveling around the world, if you like, not necessarily living in a big city, you can come and work at places like Grafana Labs. But finally, and perhaps more importantly, the reason why I think open source is the way, the best way to build software is really the kind of cathedral and bazaar model. Again, this is another popular book that describes the difference between the proprietary large organizations building software and the loose knit community building software with the proprietary large organizations being the cathedrals with their management class of the clergy, kind of dictating what people should and shouldn’t believe and the bazaar being the markets of just chaos where people just do whatever they want and kind of scratch their own itches and that chaos and autonomy and empowerment and that scratching your own itch.
(08:27): That’s how we build software in Grafana Labs. Every project that you’ve seen has come out of a need to solve a problem that we had that an engineer sitting in their home office or in a cafe or wherever they sit nowadays, like thinking, oh, I can build a piece of software that solves this problem. I can build a piece of software that automatically instruments my golang – that would be Beyla. I can build a piece of software that helps me store logs for my Kubernetes cluster without having to put them in Elastic or in some other big fully indexed system – that would be Loki. That’s how I built Loki. And so we really encourage this motion at Grafana Labs by giving engineers a week every three months to just do whatever they want. We call it our hackathons. We ask them one thing, we ask them to record like a five minute video at the end of it, self-assemble into a team, record a five minute video, and then we’ll go through some rounds of judging that result in a popular vote for the whole company.
(09:20): And the winning project gets pretty big cash prize. We really want to encourage this. And those hackathon winning projects go on to be featured at the keynote in GrafanaCON. Often. That’s how the Assistant started. That’s how the Drilldown apps that featured in the keynote last year all started. So we really want to encourage this kind of autonomy and empowerment and really give our engineers the freedom to make their own decisions about what’s best. I will say this works really well because we build software for other software engineers. I don’t know how to do this in other industries. So I’m glad I’m in this industry, but I firmly believe that’s the best way for companies to build software. And finally, I guess this wasn’t on the original slide. I kind of wanted to save this as my, one other thing is open source is our cheat code for AI.
(10:08): And again, 2025 kind of have to mention AI, but this was a relatively recent kind of realization for us. And I know Matt kind of touched on this in the keynote where we learned that these new foundation models that have come out in the last kind of six, 12 months, they were trained on the wealth of content that the community wrote and published on the internet for us. They already know how to use our software because there are thousands of blog posts on best practices for building dashboards. There are thousands of blog posts on what each individual Prometheus metric means and what the right metrics to monitor Kafka are and so on. And they’ve been trained on this information. So we didn’t have to, right? As an open source company, these foundation models know how to use our software. This is something our proprietary vendors that we compete with, they really struggle with that, right?
(10:58): They have to build their own models and they have to invest in training their own building their own data sets and training their own models to figure out how to use their software. We just use the foundation models. This is how the hackathon, sorry, I know I’m going all over the place. The hackathon that Mat and Cyril did to build the Assistant was only two months ago. This is how we were able to so quickly build such an impressive piece of technology like the Assistant and get it into the hands of our users and customers so quickly. So this was a relatively recent realization for us, but this is why open source is so important at Grafana Labs. It’s because yes, it’s how we find our customers. That only works if you don’t cannibalize too much of your community. We only have like 6,000 contracted customers, but there are tens of millions of users of Grafana.
(11:43): Yes. Open source is how we’ve disrupted like a really stale ecosystem, right? How we’ve commoditized telemetry, how we’ve prioritized interoperability. Yes, open source is the best way to build software. Something I’m very passionate about. And the culture internally within Grafana Labs is that of a loosely knit open source community. And yes, open source is the cheat code for AI, is how we’ve delivered such a cool AI system experience so quickly and how we’re hopefully going to get it in much broader hands. And I’m not going to put Mat and Cyril on the spot, but hopefully GA it in three months. Yeah, maybe. We’ll see. So that’s my talk. That’s my closing keynote.