Thank you, Scaleway.
Disclaimer: the post title was humbly inspired by Roxanne Varza :)
Hello world Online.net
The last years I passed working at Online.net were incredible, and it began with an uncommon arrival; 5 years ago, I worked with Arnaud & Solvik on Iliad (the group of Free/Free Mobile/Online.net) projects in my free time, doing side projects. We managed the streaming event of the Free Mobile launch (biggest streaming event in France, at the time), and we brainstormed a lot about designing the heart of a new kind of school, a.k.a 42. After some nights with Arnaud, CEO of Online.net, I decided to join the company.
The story behind Online.net
In 1999, Xavier Niel created the Free ISP in France so everyone can access the internet, but at that time, there weren’t so many things to surf. To equilibrate the balance, he created Online.net, formerly a web-hosting company, so everyone can also create Internet content. Free and Online.net are both driven by the technical, always innovating to create new services, and always optimizing to reduce prices. In 2006, Arnaud de Bermingham created the Dedibox dedicated servers following with the same DNA (technologically driven, high-quality services at optimized prices); Xavier quickly asked Arnaud to come at Online.net with his Dedibox and merge the two companies. Arnaud built his datacenters finally to offer all the hosting offers landscape: web hosting, dedicated servers, dedicated racks & dedicated datacenter rooms.
The day I learned this story, I fell in love with the company and promised me always to remember the group DNA each time I wanted to start something new.
Arnaud hired me without any idea of what I could do; on day 1, we walked through the office while looking for a tech team with an available slot for me. I finally joined the hardware laboratory.
Creating hardware stuff
I passed my first year as “system engineer” discovering and working on hardware development. In a team of 4 people, we designed, built, and wrote all the firmware and tools to manage a new kind of very high-density, energy-efficient, low-cost, dedicated ARM servers.
The pivot
While dreaming of remaking the world with Solvik on his sofa, I decided to suggest a pivot from the original idea of creating a new dedicated server offer (a.k.a Dedibox) and, instead, creating a Cloud offer.
We designed a 100% API-based managed hardware. We were able to control the power, the serial port, the monitoring, the GPIOs, the network, and the storage of dedicated hardware in almost the same way we usually do with virtualization, but with real hardware.
We optimized the cost with low-power consumption and a very high-density design: more than 900 nodes in a rack, with less than ten cables.
Arnaud and I crafted the business plan and the road map; we had everything to start the adventure.
From 1 to 15
We decided to create a standalone brand with its offices and methods, so we can keep the focus on our product while maintaining a close relationship with Online.net tech teams for sharing knowledge and using their technical support.
My first mission was to recruit the original team. Using my school network, and thanks to the help of Online.net, we successfully found all the 12 new profiles and three guys from Online.net in 3 weeks. four years later, 11 of the 15 first recruitments are still there. \o/
From an idea to a beta
We wrote about 100 services, workers, and APIs to handle the internal business needs and communicate with the hardware — a datacenter manager/hypervisor, but for dedicated servers and real network switches.
We needed to write a new billing system, because, at Online, we were selling a monthly servers and services subscription, and we needed a usage-based billing (similar to telecom billing).
Finally, we needed a web console that is completely static and API-based.
We opened a public beta and shortly became known in the datacenter/developer world via very cool articles from our users.
I’m satisfied with most of those choices; however, we did make some errors :)
- We tried to design the core business around a NoSQL database (Riak) and switched back to a relational one while keeping NoSQL for specific datasets.
- We underestimated the lack of upstream kernel support with our new hardware. The SoC we were using, was usually employed in NAS servers and mobile phones, which only need stability and performance, so they don’t care about having a closed-source patched kernel, but our users do.
- We took too long to open the beta to people, we didn’t get enough feedback, and our hardware was aging.
From a beta to a business
Scaleway is now a real business, with an excellent, organic growth, regularly compared with legacy competitors, and with a sharp image of innovation.
We launched more than 600.000 server sessions, and we are ordering new servers by batches of multiple thousands.
We are working on three new hardware servers and the opening of new POPs.
Our communities on GitHub, IRC, and Discourse are growing. The future looks very promising.
Thank you, Scaleway.
I recently decided to quit the company I started four years ago.
It was difficult for me to make this decision, and it wasn’t pleasant for some colleagues to learn of my choice.
But the project is in very solid hands, and that helped me a lot. I feel very lucky and happy to leave the company while staying friends with everyone, and I have the feeling that the project will expand, even without me.
Conclusion
Thank you to all my first colleagues at Scaleway, here at the very beginning with me: Arnaud Chong, Bastien Chatelard, Edouard Bonlieu, Fabrice Sodogandji, Gilles Dartiguelongue, Henley Saramandif, Julien Castets, Kévin Deldycke, Lucas Di Cioccio, Nicolas Limage, Quentin Perez, Renaud Mariana, Romain Gay, Sébastien Rannou & Yann Léger. Thank you for believing me very early on, joining the adventure, and creating with me what Scaleway is today. Thank you to all the new colleagues at Scaleway for joining our forces recently.
Thank you to all my previous colleagues at the hardware lab: Grégoire De Turckheim, Jérome Malinge, Pierre-Olivier Roumier & Vincent Auclair. The year I passed with you was the most intensive, regarding learning, and, without your servers, Scaleway would probably not exist.
Thank you to all my previous colleagues at Online.net and in the Iliad group: the network, system, and developer teams, and also the administrative, the Datacenter technicians, and the support guys. Thanks to you, I was able to keep a focus on creating things.
A special thanks to Solvik. He was the first to believe in me in this adventure. We often mixed our passions to bootstrap ideas, create POCs, and create joint projects between Online and Scaleway.
Another special thanks to Arnaud, who hired me, just based on a feeling, who gave me the opportunity to work in the hardware lab, who helped me to pitch for the project and raise the initial money to create Scaleway, who let me be very autonomous in designing one of his strategic plans, who let me order millions of Euros worth of hardware, and who stopped me before I went too far :). Thank you, also, for being very flexible when I became a dad last year.
I technically, and socially learned a lot. I keep these lessons in the memories of multiple “life achievements.”
I cultivated and developed my passion for creating things and for entrepreneurship; now, it’s time to start^Wdisrupt something new — hosted by Scaleway, of course.
Credits: Photos took on my Free Mobile’s iPhone
