Industry Spotlight: Serverfarm’s Arun Shenoy
There are many niches within the colocation and data center marketplace beyond the big REIT names that get most of the press. One company taking its own approach to the marketplace is Serverfarm, a US-based data center which operates a global footprint of both multitenant and managed enterprise data centers. With us today is Arun Shenoy, Serverfarm’s SVP of Sales & Marketing. Arun has been in the data center business since the late ‘90s, and joined Serverfarm two and a half years ago.
TR: What does Serverfarm focus on? What does your infrastructure look like today?
AS: We see ourselves as a service provider’s service provider. The vast majority of our customers are systems integrators or service providers in their own right, such as Zayo, Cyxtera, CenturyLink, and some of the big hyperscalers. We primarily own and operate infrastructure for those customers, but we also operate data center infrastructure both in the facility space and the IT space for customers where we don’t own the data centers. For instance, we manage the data centers both in facility and IT terms for a large global car manufacturer. Our footprint spans 80 locations across 33 countries with 250MW of power. One of our specialties is to take enterprise data centers, right-size them for the organization that we acquire the asset from, and lease back to them just the portion of the facility that they need. In some cases, like the global car manufacturer, we run them as single-tenant environments. For others, we reinvest in the facilities and release that capacity into the market. We have become good at spotting which enterprise facilities are ‘multi-tenant-able’, if I can make up a verb. Essentially what we do is we help people manage physical infrastructure and take the pain out of it.
TR: You recently acquired a facility in Amsterdam, what attracted you to that opportunity?
AS: That was an acquisition of a bigger portfolio of assets, and that one facility is quite large: 200,000 square feet. We can get something in the region of 23-24MW of IT power into just that one location. Since last June there has been a moratorium placed on all new data center development and new provisioning of power to data center environments inside of the Amsterdam city. Because there is no ability for someone to build a data center and request more power, we saw that as a strategic investment into capacity that is already there. It has a ton of space and power available and is hugely well-connected, which makes it a really interesting proposition for customers who need to be in the Amsterdam area but either can’t build their own or can’t find capacity in the right space. There is still plenty of capacity available in the surrounding areas, but if someone has some kind of latency requirement or a location dependency to be close to the city center, this location will be interesting to them.
TR: You also recently did a deal with 5NINES, how did that fit into your plans?
AS: 5NINES had a management contract with a large global systems integrator that we took on. We now operate facilities for that systems integrator in Canada, the US, the Nordics, UK and Ireland, Netherlands, and a couple of additional European countries. We do everything for them, including M&E projects, the IT change management environment, and facility management.
TR: What are you putting your resources toward today?
AS: Toward two or three locations. Our focus on the enterprise market continues: acquiring enterprise-class assets usually in markets that are somewhat constrained in terms of capacity. London, Frankfurt, Amsterdam, and Paris in particular, from a European standpoint. Then also in some of the key metros in the US: Chicago, Toronto, Atlanta, and Silicon Valley. Generally speaking, the focus there is as always on finding existing facilities where there is an enterprise tenant who essentially has more capacity than they therefore see a need for and we can conduct some kind of sale and leaseback transaction. That would make them a tenant in, ultimately, what will become a multitenant facility. As with all of these things, location is a key part of that commercial model.
TR: How actively are you looking for such opportunities?
AS: We generally have somewhere in the region of 15 to 20 active deals. The market is pretty fluid. Some organizations that go to market then decide that they want to kind of retrench, but then come back to the market a year later.
TR: What market trends are driving such organizations toward this option?
AS: I see organizations that maybe three or four years ago said that they were going to move to the cloud, but are now finding some of their cloud migrations are either taking much longer or are more expensive than they were expecting, or the environment might’ve changed from a regulatory perspective. And that is giving them pause to consider staying where they are. Even though we can see a ton of interest in the enterprise space to migrate to the cloud, there is still some doubt in terms of what enterprises really want to do with their portfolios.
TR: How does your InCommand technology fit into your business model?
AS: Our entire operating model is based on technology, so one of the ways that we drive efficiencies where operating infrastructure is concerned is obviously through the use of technology. About 10 or so years ago, we started really trying to look at how to make our customers as efficient as possible. One thing that was and is very clear to us when we acquire sites is that most organizations are very good at using space, but not very good at using power. We really see two key reasons for that. One is they don’t have the right tools in place that would allow them to do capacity forecasting and demand planning so that they can deploy infrastructure in the most efficient way possible. As a result, they tend to run out of space much more quickly than they run out of power. Then the other part is process. In an enterprise organization, you will frequently have multiple parts of the organization making decisions about where capacity should be deployed, but there isn’t a single planning function and therefore there is a lack of ability to create efficiencies. So we developed our InCommand platform, which brings together all of the facility and IT physical infrastructure: real time information from building management systems, real time information from IT & FM environments, etc., into a central planning functioning where we make decisions about where and how infrastructure needs to be deployed and monitored. We use InCommand for everything from maintenance to asset management to ticketing to work order creation to capacity planning to demand forecasting, to reporting, to dashboards.
TR: And how do your customers interact with it?
AS: For our colo customers, it is basically their eyes and ears into our environment, their ability to look into their spaces and see exactly what’s going on: power utilization, temperature or humidity performance, capacity planning, etc. But it is also the way that they interact with us, to request things like smart hands, remote hands, access to the building, badging or notification of maintenance events. For organizations where we deliver it as a fully managed service outside of our environment, we’re doing everything including providing people for facility management and/or IT change management functions.
TR: Are you still adding capabilities to InCommand?
AS: Yes, all the time. It’s mature in the sense that we use it to run all of our critical infrastructure and have done for a long time, but at the same time it’s also developing. We are developing things like API functionality for integration with more third party systems and improving usability from a features perspective. But it’s important to realize that we don’t market InCommand as a software platform. We’ve always had the very strong belief that what customers are looking for is not another tool to manage or another piece of technology; what they want is a partner that can use technology to give back to them a managed service that gives them a better outcome. That could be greater efficiencies of space and power, greater utilization of capital, lower costs of running, or lower costs from a fixed overhead perspective.
TR: In what unique ways does Serverfarm approach the question of sustainability and embedded carbon footprints?
AS: We obviously feel very passionately about it, which by the way is not unique as so do lots of other people. But I think the important thing for us is really developing and creating some real efficiency from an embodied carbon perspective on behalf of our customers. We like to acquire an existing facility, which may or may not be a data center, and convert or refresh it, as we did in the Chicago market, or as we did in London with the acquisition of the T-Systems site. When we do that, we’re creating capacity in a market without the need of having to build a new facility. So for example, some of the analysis that we did in Chicago shows one customer with a one megawatt load saves four years’ worth of embodied carbon, both in terms of construction and in terms of operation, as compared to a new facility. That is a big number, we’re not 10% better off, but rather 88% better off compared to a new facility. Of course, we are in the same space as all of the other leading companies in terms of low PUE and sourcing of electricity from green sources. In London, for example, we refurbished the cooling infrastructure using free-cooling chillers to drive PUEs much lower than they have been with the previous use of the facility.
TR: How has the COVID-19 pandemic affected your operations?
AS: The first thing, which we welcomed pretty much everywhere that we operate, was a recognition by the local authorities that data center staff were classified as key workers. That included construction as well, and since we were in the middle of constructing our Toronto facility, in the end COVID-19 cost us less than a month in terms of schedule. We managed to very quickly figure out what we needed to for personal protective environments and social distancing. The good thing about data center environments is they are already quite well socially distanced in terms of staffing. You don’t tend to get a huge concentration of people sitting or standing in a small environment. But we very quickly recognized that what we wanted to do was to limit traffic in and out. We communicated with our customers very early on, and we dropped our remote hands pricing significantly the first week that most of the lockdowns took place. Others in the industry were perhaps a bit more sort of opportunistic, but we took a different view. We didn’t think that was helpful. Reducing our remote hands pricing was going to help our staff as much as it would help our customer’s staff and just made it a little bit easier for customers to consume the resources that we have on-site. We haven’t had any severe cases, although we have had a couple of people that have showed symptoms, were tested, and have come back negative. I think it helps that we’re not a retail colo, and we don’t have hundreds of people coming in and out of a facility in any case.
TR: What other effects have you seen in doing business in this environment across your ecosystem?
AS: I guess we’ve all had to deal with becoming kind of experts at using all of the various kind of video conferencing tools. But I think we have actually become closer to our customers than ever, partly because we’re all collectively much more available. It’s much easier to find time in someone’s calendar than it used to be because no one’s traveling. I think we have also found more about the personal side of our customers’ lives than I think we ever would have done. We are dialing into people’s homes, hearing kids running around and all that. In an office environment that conversation doesn’t come up as much. There’s this informality now with the current environment in terms of how people are dealing with each other.
But I think there are also a few downsides is that, just from a well-being perspective, people are actually finding it very hard to turn off. You don’t have “I’m now leaving the office. I have now arrived at home.”
TR: How do you think the industry as a whole has held up under the strains put on it this year?
AS: I do have to say that I think we have done amazingly well. Organizations that have come together, collaborated, and figured out that they need to reshape some of their networks. We saw a lot of our bigger customers move demand into our facilities much faster than we were expecting. Utilization rates, from a power perspective, went up as they started to move workloads into our environments. They might have had hardware provisioned that they were maybe intending to light up in six or eight months from now that they pulled forward. The biggest challenge actually has been for enterprise IT organizations, some of whom have had to completely rethink their network security models. They had figured out this very well-secured enterprise network environment, and then suddenly, overnight, everyone was at home on unsecured networks with VPN deployments not as broadly kind of deployed as they might have intended. Those are the people that have done the most amazing job: the ones who haven’t had a solution that was ready for them to just deploy. They have had to go fix problems that they probably haven’t thought about having to fix maybe even ever. Most organizations never envisaged that almost 100% of their workers would suddenly be remote.
TR: Thank you for talking with Telecom Ramblings!