From sheet metal to connected server (1)

With my teams, we manage the infrastructures that support OVH. We design and deploy the datacenters, the servers and the network. Through this series of blog articles, you can join me in discovering the "heart of the matrix", and get a better understanding of how our model works. I’ll start by telling you the story of a guy from the North of France, who went to work in Silicon Valley before joining OVH.

I arrived at OVH a year ago, in July 2017. I’d worked at Google for 12 years, including 5 years at its headquarters in Silicon Valley, developing infrastructures for the world leader in search engines. My passion is fiber networks, datacenters, machines, and energy on an industrial scale.


From Mountain View to Roubaix Valley

You may be wondering what could entice a Global Infrastructure Director from Google to switch to OVH. To be honest, if anyone had told me that one day I’d return to the North of France, I wouldn't have believed it. I’m a native of this region. A "techie" who went to California to make his dream come true. And in 2005, I’d made it. I was working at Google headquarters. But at the same time, I was keeping an eye on the adventures of OVH and Octave, following all of the public notices on the main website as well as the status tasks. You could say that I have the OVH vibe, that I am "OVH-compatible". In 2005, as I was helping Google choose its fiber optic equipment, I realized that Octave was making the same choices. From Silicon Valley, I watched OVH grow and become a major player in the digital industry. The company was stating its ambitions to become a European leader in cloud services. So when the opportunity presented itself, I took a closer look.


Integration to the max

But it only was once I had done a complete tour of the OVH site in Roubaix that I realized the sheer level of integration, from server production to datacenter connectivity. That's when I truly realized what OVH was all about, as well as the scope of what Octave had already built. Everything was there, concentrated in one place: design, deployment, datacenter management, server assembly in France and Canada, and the development and operation of its own backbone network. Integration is pushed to the extreme — I've never seen anything like it anywhere else.  We also have the know-how to build our own server racks on-premises from sheet metal that we cut to size, using our equipment. This first visit to Roubaix was decisive for me. I immediately understood the benefits of an organization like this for our customers. It’s well-known that OVH builds and operates its own datacenters. But not many people know that OVH assembles its own machines, and that we control the entire production line, from bare sheet metal all the way to the connected servers in the datacenter. That is where our strength lies.

The first advantage of our model is obvious: we’re able to reduce costs from A to Z. By having full control over all elements of the chain, we can choose and test all of our hardware down to the smallest components, achieving considerable economies of scale as a result. But above all, having such a high level of control also makes us much more agile.

When a customer asks for higher performance on their server, requests more memory or a higher computing capacity, we can improve things in real time. The reason we’re able to do this is because our datacenter technicians can intervene directly on the server. We don't have to wait for a service provider to schedule an intervention. This is one thing that sets us apart from most of our competitors. Our responsiveness is unrivalled.

The same principle applies to server customization. No other provider offers such a wide range of server models, and the reason for this is that they simply can’t do it as swiftly as OVH.

This vertical integration has also allowed us to push the automation of logistics processes to an utmost degree, from provisioning server components to delivering services to customers.  Everything has been custom designed, from the hardware to the software, which allows us to deliver machines in just 120 seconds.

We have a capacity for increased, rapid and intelligent innovation at all levels of the value chain, for each of its components. For instance, we use horizontal racks, whereas the entire industry uses vertical models. Why? Because they are much easier to move around on motorized trolleys in the datacenter. Innovation is driven by all the business lines and by the organization itself, via processes that enable everyone to submit their project. If an employee's idea for improvement is validated, it is immediately deployed. We can also test ideas, even if they are unusual. Water-cooling was one of them, back in 2002. Everywhere, dialogue is encouraged between all parts of the company, resulting in a colossal gain in efficiency.


Industrialization on an international scale

OVH’s second major asset is its ability to replicate its model quickly in markets that are very different from each other. The deployment model is also unique. Firstly, all the skills already exist in-house, and the technologies have already been proven. All we need to do is to buy an existing building near an OVH network connection, renovate it, and transform it into a datacenter. First of all, the servers are tested and preconfigured at the factory. They are then delivered in racks, the horizontal racks mentioned earlier, ready to be connected. All we have to do is connect the electrical cables and the water-cooling network.

For the record, when I arrived, I thought that my job title should have been Chief Infrastructure Officer, as was the case in most similar companies. But as I got to know OVH better, I realized how important the industrial aspect is for the company.

Our new server rack model makes us highly agile, and plays a part in setting our datacenters apart from any others.

Compared with many other datacenter operators who initially deploy large buildings, the OVH model instead provides a high degree of modularity, and allows small datacenters to be deployed.  By adopting this approach we can test new locations, make decisions, and evolve in line with the local market. And in capital terms, the datacenter can generate its first R.O.I. in a few months, which is an extremely fast turnaround.

On a personal level, working across the entire chain on such a scale represents an opportunity that is both unique and incredibly interesting. And the fact that a French company could do this out of Roubaix was just too tempting an opportunity to miss for a physical-infrastructure geek like me.


What about the human factor?

There is also a spirit, the fabled OVH mindset . The company history is rooted in the local employment area, which was hit hard by the crisis. Our staff show a level of company loyalty that is rarely seen elsewhere.  The mindset isn’t volatile like in California, where people are only there to advance their careers. Here, there’s a common goal and a family spirit, which is marked by conviviality and closeness. And this is even more true in the industrial sector. Employees are proud of their work and what the company accomplishes, as shown through our 95% commitment rate. There is no formal hierarchy, relationships between people are very direct, and we all work as one team. I was impressed by the solidarity between different departments.

To continue benefitting from all the intelligence deployed in the infrastructure, we need to meet a number of challenges, due to the fact that we have grown very quickly: we need to strengthen our foundations, improve our production environment, and manage the obsolescence of our infrastructure. For server production, we’re in the process of upgrading to factory 4.0. During the summer, we’ll move production from Roubaix to our new premises in Croix, which is still in the Lille metropolis. We’re using this move as an opportunity to equip ourselves with new state-of-the-art equipment, particularly in terms of inventory management as well as new workstations. As for the datacenters, following the lessons we learned from the incidents of last fall, we have implemented a thorough action plan. And, following the launch of OVH in the US, we are following a well-thought-out deployment plan in two modes: "growth on demand" and "Datacenter and Network as-a-service".

At OVH, my ambition is to contribute to building a world-class, innovative and agile digital platform, while remaining efficient from an economic and energy point of view. Like this, we can maintain our services in a transparent and virtually seamless manner for our customers. The next articles will go into more detail about our server production, datacenters and innovations.