insights! #96: How Local Zones are redefining the rules of internationalisation!

Joubin Rahimi:

Great to have you back for a new episode of Insights. Today the third episode with Falk from OVHCloud. Welcome, glad you're here with us, and the topic is internationalisation.

Falk Weinreich:

Thank you. Thanks for the invitation for the third time. I'm a regular now.

Joubin Rahimi:

Total.

Falk Weinreich:

I feel at home here with you.

Joubin Rahimi:

We'll make ourselves a name badge.

Falk Weinreich:

Name badge, yes, on the sofa, exactly. Yes, internationalisation, that's a good keyword. According to studies today, IDC, but also ISG, there are several providers, data analysts, who also always track this, in the data centre environment where we also have assets in various countries, we're already represented today in over 30 data centres worldwide, from Singapore, Australia, Canada, USA, Europe. We already cover the most important markets today. But we don't want to stop there. We want to become even more global. And to that end we've launched a new product. Not brand new, that was now already three, four, five months ago, but we call it "Local Zone". That means that our public cloud products, we're not doing this for all products, because technologically that would mean too much effort, but for the core products, especially public cloud, we're now going more granular, rolling out smaller offerings worldwide faster. Worldwide means really, we've just launched Rabat, three, four months ago Zurich, in December Vienna is next, we've never had services in Madrid before. We were never in the Netherlands before. We'll continue expanding further into Asia. We plan to launch over 150 zones in the next two years, in almost 50, 60 countries and thereby make the lead over our competition even clearer, when it comes to global availability, to deliver that, with the main advantage for the customer of course being that they can choose a provider regardless of where they are, almost regardless of which country they're in. Some customers still find that important. Sovereignty, data sovereignty doesn't necessarily have to do with geography, but for other reasons that does exist, especially Switzerland. I like to use that as an example: Swiss customers really want that. The service should please also be located in Switzerland, alongside data sovereignty. But then also to have a provider who can offer Switzerland, but also Spain, Italy, Canada, Brazil, everywhere in the world. That's especially in our region too, Switzerland, Germany above all, the Mittelstand is an export nation. We're an export nation and customers expect us to take them along on the journey worldwide as well. So availability worldwide, the sovereignty topics and also latency, because of course you're close to the customer.

Joubin Rahimi:

That's an exciting point, because in the preliminary talk we'd said: 'What actually interests customers? That's your topic, isn't it.' But the exciting thing, we found out afterwards: we have customers in the aviation industry, we have customers in other industry sectors, who need very short latencies and need large data centres, data centre power, which they then don't have where they need it. I found that exciting. So predictive maintenance was a topic, other analyses too. And it only became clear to me afterwards that they'd said: 'That's great, because my plane, my ship, my car is out there somewhere and I need a mega-short latency, and that's already important: do I have 8 milliseconds or do I have 800 milliseconds?'

Falk Weinreich:

Yes, ultimately it’s like a layered system, like certain applications, cold storage, archives, whatever – that can genuinely be anywhere in the world where it’s efficient, has cheap electricity, and I very, very rarely, hopefully never, have to touch it. And then comes the next layer for certain compute performance that really needs super efficiency right now, and there the licence isn’t quite as important to the customer yet either. And then there are certain applications where it’s simply so critical that it can’t take long. We see that in gaming, funnily enough.

Joubin Rahimi:

Total.

Falk Weinreich:

If the player takes too long, he's finished. Accordingly however-

Joubin Rahimi:

Doors get wrecked too, and mice, when the kids then break the things because the latency was too slow.

Falk Weinreich:

Well, in many areas latency is really important, and being able to offer that from a single source, so the customer can basically expand their portfolio with us, so they say: "I can entrust you with even more of my data. I like you anyway because you're data-sovereign. I like you because you don't use proprietary systems, because you're technology-open. I like you because there's no lock-in effect and no egress fees for traffic." There are certain reasons why customers have stayed loyal to us for a very long time, and we can then take those customers along on the journey, for certain applications, to say: "Now you can also use us in, I'll call it Rabat now, precisely because we've just launched it, if you need us on-site there, if you need a cloud there, if you want to reach your customers there, because that's simply your market, then we can now serve you there too, and soon in Brazil, soon in lots of countries in Asia. My region is of course more Switzerland and Austria and Germany naturally, but as I said at the start, German companies are global exporters. We're an export nation. That means it's also really important for my customers here that we now have a location in Amsterdam, and in Brussels, and in all kinds of countries, because we want to take our customers along on this journey, and customers have also told us this in feedback conversations. The theme of our product development is always to be close to the customer. We don't do this from our own chamber or the Eiffel Tower. Ivory tower, for us it's the French Eiffel Tower here. Right, Freudian slip there. But it really is the case, we develop this together with the customer, with their needs, and that's gone down pretty well. So we're seeing an extreme rise in the number of customers. Also, because it's a Local Zone, it's not the customers' main data represented there — so these aren't applications that go completely over the top, more like smaller-scale revenues too, for many customers, or one customer with many locations at the same time. What's important to know though is the journey there — if we then find in a country that this is growing so much that this Local Zone really needs to be expanded, also to other products, that we bring our entire portfolio from Private Cloud, Bare Metal, then also into that market, then we have this option now, because we're entering new markets through a relatively CapEx-friendly investment, and once they reach that size, then we can add on and offer the traditional, larger data centres there too. But of course it's also a capital efficiency question.

Joubin Rahimi:

How expensive is a data centre anyway?

Falk Weinreich:

Well, that's funny, it obviously always varies, but if you look at the media today, and we're part of the market, so I'm quoting more the general view here, and when Google, Microsoft and AWS offer their 100-megawatt data centres in these markets, of course that's several million, sometimes 100 million, that needs to be spent, raised as capital, to build a huge new data centre like that. Ours are a bit more efficient in terms of capital needs, because we typically don't go looking in, say, downtown Frankfurt. Ours is in Limburg, where the real estate market is different. So we have lower costs, just in terms of the land alone. And traditionally, we build brownfield. Meaning we're sustainable too. The ecological footprint is something we pay attention to, using existing industrial sites wherever possible. We — brackets open — we can do that — brackets close — because we cool with water. We covered that in our first episode. We don't need the air circulation in the data centre the way maybe more modern data centres do, because we don't circulate air, we work purely with water. That means we can happily move into old industrial sites too. Means no excavation. We don't have to first dig up huge amounts of dust and old materials and dump them in landfills, we can use what's already there. And that naturally saves capital too. So we're a bit more efficient. We don't need quite as much money as the American hyperscalers, but still, the first step is to be faster and more localised in the local zone, then scale up in the markets as needed.

Joubin Rahimi:

Great. Many thanks for the insight. I think time's up again for us to post it on LinkedIn. So if you still have questions about Local Zones or which cases you typically implement there, feel free to drop them in the comments below. We or our teams are very happy to answer. Thanks, Falk.

Falk Weinreich:

Great. Thanks, Joubin.