Email a colleague    

June 2018

At the ROOT of Wholesale Data Center Value: Fast Deployments, Energy Efficiency, and the Waterfalls of Quebec

At the ROOT of Wholesale Data Center Value: Fast Deployments, Energy Efficiency, and the Waterfalls of Quebec

Every successful wholesale data center is driven by special qualities that create a magnetic force to attract customers:

  • For DataGryd, it’s being the only large footprint build-to-suit data center in Manhattan, the epicenter of America’s financial, advertising, and mass media brainpower.
  • For EdgeConneX, it’s being the gateway for cable giants such as Comcast, with their high-speed internet pipes streaming 70 to 80% of content to American homes.
  • For Equinix’s many carrier hotels in Ashburn, it’s the power of the U.S.  Federal government and the thousands of communications, defense, and systems integration firms who live off the Washington, D.C. beltway.
  • And for 1025Connect, it’s being a preferred landing site for Atlantic undersea cables that connect America and Europe.

Now, just at the point when you think the data center business has matured and incumbents are sitting pretty, along comes an innovator who shakes up the status quo and makes you rethink your assumptions.

Case in point is Montreal’s ROOT Data Center — in business nearly five years.  ROOT’s success is drawing hyperscaler customers — and causing many rivals’ heads to swivel North.  I met ROOT’s President and CEO, AJ Byers, at the ITW show where he explained in compelling detail:

  • Why Montreal is fast-becoming a data hub for all of Canada;
  • How ROOT’s construction speed and innovative engineering has won business from many of the Tier 1 cloud hyperscalers; and
  • Why latency delays from Canada will not impair ROOT’s ability to serve 98% of business applications in cities of the U.S.  Northeast.

Dan Baker, Editor, Top Operator: AJ, I understand your business is deliberately tilted to attracting large wholesaler customers.

AJ Byers:: Exactly; our business premise is that data center operators had become very retail- and enterprise-space focused.  Hardly anyone is really looking at the market from a wholesale perspective.

Wholesaler customers need an ability to build capacity very quickly.  The hyperscale cloud providers, for example, not only need large capacity: they also want to be sure a data center can keep up with its growth demands.

This is why we devised construction techniques inside ROOT which allow us to deploy large capacity volumes, say 3 megawatts, in only 90 to 120 days from contract sign.  That’s very fast in an industry that usually takes 10 to 12 months to build that capacity.

Who are your actual customers?

Number one, it’s the hyperscale cloud providers.  All the cloud guys are coming to Montreal.  Microsoft, Google and Amazon have all launched in Quebec.  They plan to initially service the entire Canadian region from Montreal — a major deployment in the country.

After the hyperscale companies, we also serve some of the largest tech companies in the world.  Then there’s enterprise Canada.  Even mid-market Canada will buy from us because they love the idea of leveraging the low deployment/operating costs we have built into the wholesale platform.

Why are these companies coming to you?

I think the key differentiators are our 1) rapid construction; 2) super high energy efficiency; 3) environmental friendliness; and 4) low cost — thanks to our low electricity price and ability to build out very efficiently.

On top of these factors are the table stakes requirements that every data center provider worries about: high security and 100% reliability.

Not so obvious are three engineering practices at ROOT that set us apart:

  • In-House Design — We have a very innovative approach to in-house engineering and design.  We are constantly asking ourselves: “How can we modify our existing design to further improve our time-to-deploy and cost effectiveness?”
  • Design-In Easy-to-Source Components — We are careful not to over-design.  For instance, for our 3 megawatt pod design, we choose 1 megawatt generators because those generators are readily available, and we can deploy faster.
  • Supplier Inventory — Finally, we make sure our suppliers maintain a ready stock on the components we need, so they are readily available.

What about redundancy?

We have two centers in Montreal to allow our customers to do a redundant solution in the city.  Three to four fiber providers are located in each center.  And at each center there are 6 to 10 carriers to connect with today.

On top of that, we are talking to all of the cloud connectivity/SDN companies.  We haven’t picked a provider yet, but that’s one reason we are here at ITW — to chat with all of the cloud connectivity players.

So far, cloud connect in Canada has not really taken off, but demand up in Canada is definitely increasing.  So we fully expect to deploy cloud connect technology in the next six months.

Your business model reminds me of EdgeConneX’s strategy of deploying first for Comcast and letting gravity pull other providers in.

Yes, in that sense, EdgeConneX’s model is similar since large-scale wholesale customers are the cornerstone of what we do.  And because these are large deployments, achieving low cost and low PUE are big factors.

PUE ultimately determines what the customers’ cost is going to be on top of the cost of space alone.  Our designed PUE is 1.17 in a market where the norm is 1.3 to 1.4.  This is a very efficient design that uses the superior KyotoCooling technology; we are also doing our own custom design on our hot aisle containment.  Our custom-designed cabinets also increase air flow.  Our aim is to drive extremely efficient design in the center.

Another big asset for us is Quebec itself, where 99.99% of the electricity is renewable.  The hyperscalers  love renewable energy, and Quebec is powered by hydro-electric plants.

Quebec also has the lowest cost of electricity in all of North America — a lot of that power comes from dams in the northwest regions of Quebec.

We have spoken to many U.S.-based companies and they are finding that by deploying in Quebec, they can service all the U.S. Northeast and do it more economically.

But what about the issue of latency?  The data center experts I’ve spoken to say the hyperscale firms are looking for extremely low latency.

If you look at the path from Montreal to Virginia, that’s about 8 to 10 milliseconds of latency.  So, for 98% of applications, that is not a problem.

It would be harder to serve San Francisco from Quebec, but I can serve all the Northeast U.S.  For many applications, you don’t generally need to be in the city itself.

But can you support real-time applications such as video?

If you can wait 30 milliseconds for your video to start, you can, and hardly anyone will notice the difference.

Video applications buffer their traffic; there’s a latency delay while the application figures out how fast the packets are going to arrive.  If the buffer fills in less than half a second, the video will start in less than half a second.

The other side of the latency, of course, is proper local caching and content delivery.  For example, if thousands of people are watching a video in Chicago, it makes no sense to pull the video a thousand times from New York.  You access that video once and deliver it a thousand times over a short distance using multiple short pipes in Chicago.

Latency impacts are very application-specific, and this is why I say that in 98% of business applications, there are no latency or location issues.  The originating content could be anywhere in the country.  You can service all of Canada from Montreal; you can service half of U.S. from Montreal as well.

Even still, there’s great variety in compute applications.  Some apps require you to be on the edge; some of them need to be central; and some of them you could put anywhere.  I’m the first to say there are use cases for data centers to be in multiple different locations.  There’s no right answer.  Different applications require different answers.

What markets can you reach out of Montreal?

Montreal is uniquely positioned.  The city has quickly gone from almost no data centers three years ago to being the largest data center hub in Canada.

To give you a bit more data about Canada: 80% of businesses are within 400 kilometers of Montreal, so serving all of Ontario and Quebec from Montreal gives you 80% of the Canadian market.

The other 20% of the market is spread across the eastern Maritime provinces, the prairies, and British Columbia.  That remaining 20% is highly spread out.  Most people focus on that 80%.  This is why Montreal has become the biggest data center hub in Canada.

What about future plans?  Will you deploy in other markets outside of Montreal?

We are not interested in entering Tier 1 markets.  Some great data center providers are already there.  For us, it is about taking all the companies who came to Canada and helping them reach other Tier 2 markets around the world.

For example, if they are going to South American markets, I think we would replicate our platform there.

Longer term, we will probably go to many smaller countries because of data center residency requirements.  If you want to do business with large enterprises or banking or governments in any country, you need to have a presence in that country to keep the data in the country.

The cloud companies will eventually go into every country with a good-sized GDP.  So for us, it’s about looking at other countries around the world who have decent GDPs and decent economies where a hyperscale cloud presence is not there yet.

Can you tell us bit more about your personal background in the business?

I’ve been in the data center industry for 20 years.  We started a company in 1996 that built the first independent and carrier-neutral data center in Canada.  Our value proposition was, “Would you rather deal with a telco or deal with a real data center company?”

That business was ultimately sold to Primus Canada.  And inside of Primus, I built nine data center platforms and much of their telecom business as well.

Then in 2009, Primus sought bankruptcy protection and decided to sell off all its assets.  When they sold the data center business to Rogers, I became President of Rogers data centers in Canada.  I bought two more companies, merged them into Rogers, and we had 15 data centers across the country, becoming the second-largest data center platform in Canada.

I left Rogers around 2014 and I started at ROOT in 2015.

AJ, congratulations on your data center innovations.  Hearing your history, it’s pretty obvious how your experience led to ROOT’s great success.

Thanks, Dan.  I can’t over-emphasize the importance of speed.  When the hyperscalers saw how fast we could build, that got their attention.

For many of the customers we have, the number one reason they buy from us is to deploy capacity quickly.  So that’s our constant mission at ROOT, to jump-start our own design innovation to keep up with the hyperscale capacities, visions, and ambitions of our customers.

Copyright 2018 Top Operator Journal


About the Experts

AJ Byers

AJ Byers

With 20 years of experience in the data center industry, AJ has proven skills in business transformation and growth.  Recently, as president of Rogers Data Centers he led the team in the development of one of Canada’s largest data center service companies with 15 centers nationally.

As executive VP of Primus, he guided its transition from a legacy telecom provider to a full solution technology services organization.  He has also long been a pioneering force in the industry.  As COO at Magma Communications, he was instrumental in building one of Canada’s first internet data centers.

Recent Articles