To say colocation data centers are constantly evolving may be the understatement of the year. However, in which ways are colocation providers changing?
What competitive market forces and opportunities are on the front burner for both colo providers and the clients who entrust and outsource their digital livelihoods?
At this year’s fall Data Center World conference at the Ernest N. Morial Convention Center in New Orleans, I attended an excellent panel discussion, “Colocation Strategies – Questions Answered by the Top Tier Colocation Providers,” as part of the conference’s Facilities Operations Management (FOM) track.
The panel was moderated by industry analyst Liz Cruz of IHS and included:
As a starting point, Cruz asked panelists to define colocation.
Brueggeman from CyrusOne kicked off the session by explaining how colocation means “outsourcing your data center to a building that’s not owned by [your] company. Having your own building no longer provides a strategic benefit. It’s eight times more expensive to build and run your own data center, so more companies are moving to colocation as compared to do-it-yourself. Companies no longer have two- to two-and-a-half years to build their own data center when they can move into colocation that provides immediate power, space, and cooling.”
In addition to cost and speed benefits, Carey of Keystone NAP highlighted how colocation data centers provide “high-density power and redundant street power to solve application requirements. Colocation providers can do this better than companies do alone.”
Next, Cruz asked about addressing requirements (in the context of how colocation has never been one size fits all) because everyone has different power, cooling, and density requirements.
Leonard chimed in first– with a quick mention of his upcoming educational session– on the conference's final day, addressing flexibility in data centers. Speaking on behalf of ViaWest, Leonard’s more central point was “data centers are shifting from where they used to be. Now they’re built for how they’re going to be used. This is and will be the biggest sea change in data centers over the next few years. If you don’t build and operate data centers on a regular basis, you shouldn’t be in that business.”
Dovetailing on Leonard’s comments with a more capex vs. opex angle, Brueggeman compared how you can invest $30 million to build a data center or invest $30 million in capital for other things in your business. “Building your own data center has no economic benefit to your business. If you have concerns about privacy, security, or control [when deciding to outsource to a colocation provider], know that 200 Fortune 1000 companies have all overcome those concerns.”
Representing Vantage Data Centers, Lim took a more consultative approach and explained how “there’s not a one size fits all anymore. It’s more of a progression of how to make the move to a data center, so we work with clients earlier in their process. Their power requirements can be all over the place, but there is lots of choices now. The question is: ‘How far do you go to be flexible vs. how to deliver quickly?’ Clients want to react quickly.”
On addressing client requirements, Keystone NAP’s Carey explained how clients' interests are beyond just the colocation facility. Clients want “expandable capacity right inside of the data center. So what other services besides colocation do you provide? For example, designing and managing networks to connect out to the WAN to further expand your service platform delivered to clients.”
Next up, Cruz asked how many are using public cloud services.
With almost a Darwinian pragmatism, Carey explained how the public cloud is “an inevitable reality that’s going to happen. Embrace it. Facilitate connectivity to public clouds. It shouldn’t be looked at as a competitive threat. Many of you have AWS (Amazon Web Services) running in your company, and you don’t know it. So what makes sense to move to the public cloud vs. what doesn’t? Embrace public and private clouds. We also have several cloud service providers in our data center. They’re ending up in a data center somewhere, looking for speed and flexibility.”
Cruz followed this up by asking, “With colocation growing rapidly, who are the customers today vs. five years ago?
“Most surprising, the Amazon ‘graduates’ - born in the cloud, or grew up in the cloud, and they’re now realizing that public cloud is too expensive,” said Leonard. “They’re not moving to a data center of their own. They’re not building a data center. They’re moving to a colocation provider. It’s not true that you don't come back once you’ve gone to AWS or Azure. They do come back.”
When explaining how Vantage Data Centers sees its positioning as compared to public cloud providers, Lim recounted how clients are “realizing that some public cloud gets really expensive. Some companies are realizing that this is too critical, and they really need to run their own data center, among the biggest providers in IT. There’s also a resurgence of providers that used to build their own– where’s is so expensive to build your own– and the corporation built a beautiful data center that’s severely underutilized.”
Next, the public cloud conversation went more in the direction of how clients compare or ought to compare a DIY approach vs. outsourced colocation.
Companies building their own data centers are “either very underutilized or overutilized,” Brueggeman added, “With colocation, you take up chunks of power and space as you need them, to have that flexibility.”
Moreover, cost-savings are a big issue as well. “You spend $15 million per megawatt vs. how we’re spending $6 million per megawatt. Plus, you’re committing 20 years to something that you’re not excellent at.”
Besides the utilization and cost-savings, Carey from Keystone NAP added some vertical market context, “In the healthcare space, traditional hospitals had real estate. We see hospitals re-platforming their systems to move out of DIY data centers and seek multiple colocation data centers to drive their platform.”
Change is slow, and many enterprises are paying a small fortune for avoidable expenses. Lim, in particular, mentioned: “Overall enterprise [change] is still slow. There are still too many data centers. When I talk to people, they concede that I have to figure this out. We hear horror stories of people paying rent for two years because they didn’t know how to move assets because the assets sat there so long.”
Brueggeman observes, “The number one reason for colocation is disaster recovery (DR). The number two reason is that clients just want to get out of the asset business because their data center is not as efficient as they thought. For example, the Chicago Mercantile Exchange went through this. CyrusOne has been involved in a number of sale-leaseback opportunities where we’re told, ‘We don’t want to run a data center anymore,’ so it’s an opportunity to buy core assets at really good prices where there’s often very low utilization for existing clients.”
And for the benefit of those in the audience who may feel like a move to a colocation provider would be overwhelming, Carey added that " solution providers can help with the transition from DIY data center to a colocation data center. There are firms out there that are really good at project management that can physically move assets on certified safe trucks.”
Cruz’s next question: “When should someone build and not do colo?” sparked the most audience controversy, while the four panelists were largely aligned.
What sparked the controversy? A few enterprise data center managers and end users were determined to get one or more of the panelists to admit colo is not always the solution.
My observation: the panel was unflappable, and the enterprise IT managers were frustrated they could not win a moral victory; there would be no public confessions or concessions by colocation executives on this panel.
Leonard kicked off the segment by recommending you “need to evaluate the cost of DIY vs. a colocation provider. Are you big enough to commit to 20 years of operations if you're from an enterprise? Are you building a number of data centers across the world? If you do this, your first one [data center] is terrible, but the fifth one will be pretty good.”
Brueggeman added how his team could “build to suit in three months. 250K square feet in six months. So for DIY, do you have that kind of footprint? Do you have full control of the building?”
Lim– circling back to the public cloud question raised earlier– reminded the audience there is “extreme value in public cloud early on when you don’t know your workload. How can you grow vs. starting out? So the decision comes down to whether I build or not build?”
Carey also had an interesting take on when not to build your own data center. Simply ask: “What does the application stack require? And are there reasons to be in certain environments?”
From the ‘if-you-blink-you’ll-miss-it’ department, Brueggeman tipped his hat once to the build-your-own, DIY, enterprise data center manager, mentioning how he sat in on the Data Center session with Luis Zepeda of Walmart Technology [BC/DR Shifts in the Face of Global Enterprise Cloud Critical Facilities (CASE STUDY)]. The session was about “delivering computer resources to the business in the best possible way. IT infrastructure + applications + business value together.”
So perhaps if you are Walmart with 11,000+ locations worldwide, $400+ billion in annual revenue, and over 2 million employees worldwide, perhaps colocation is not the answer.
After covering the topic when colocation is not the right answer, the discussion turned toward the speed at which sales cycles and ramp-ups take place.
Recounting a recent Super Bowl Sunday when he took a call from a potential client (presumably, the potential client was not a football fan, or was intensely concerned about data center infrastructure), Keystone NAP’s Carey relayed how his team “responds very fast to potential clients– even during Super Bowl Sunday. It takes six to nine months to lay out the details of the transition. We can get through contracts very quickly, but the hard work begins after that; how do we continue delivering customer service? Is there an existing DR (disaster recovery) plan in place before moving to production workloads?”
Brueggeman added, “We’re not slowing anyone down. It’s just a matter of ensuring we’re checking all the boxes.”
Echoing a theme covered in my own Data Center World session on how buyer behaviors have changed so dramatically in recent years (“How Data Centers Use Thought Leadership To Attract World-Class Clients And Talent”), Lim explained, “the decision is already made before they even talk to us, so it becomes a process discussion. Our deals are 1 to 3-megawatt deals and take 12 to 18 months from start to finish (depending on the level of customization) and are delivered in phases.”
Leonard from ViaWest wrapped up this section by talking about his team's “faster turn, around four months, because of smaller retail customers.”
Next, recognizing colocation outsourcing as very much a partnership or marriage between the colocation provider, moderator Cruz from IHS asked: “What do you expect from your customers?”
Echoing a power-centric focus mentioned just a few minutes earlier, Carey from Keystone NAP advised clients “need to come to the table with current and future power requirements.”
Lim from Vantage Data Centers wants input on the design. “… [Besides power, you want to be] Designing your density as well. We want to be involved in design; what’s our design in the distribution? And we have some California-specific requirements. Customers are very sophisticated with very specific ideas and requirements.”
From ViaWest’s seat on the panel, Leonard described how “the biggest failure as an industry [the data center industry] is assuming customers know what to ask for. Because you don’t know, you oversubscribe, overbuy, and end up with 26% utilization of what you buy. The smaller guy makes mistakes. We have to take this out of play and paralyze IT with fear that they’re making the mistake. We only have 1,500 customers. The number of ones that absolutely know what they’re using is tremendously low. They’re not doing monitoring at the rack level. It’s so refreshing when someone comes in and knows their rack utilization. 95% don’t know.”
Brueggeman had some very similar recommendations to potential clients of CyrusOne, “Whether it’s one rack or the full building, the more sophisticated buyers often know what they need, but many times they’re off the mark. Our sales engineers work to make sure you’re not buying more power or space than you need. So be realistic and be open about what you don’t know. The last thing you want to hear is ‘You guys made us buy way too much power, and we’re not using it. So take it back.’ Our sales engineers work hard to combat that.”
So what do you say to clients/potential clients who tell you: “We don’t like colo because we can’t control it,” Cruz asked panelists, “How much control do you maintain,”?
For Leonard of ViaWest, the information people depend on or assumptions they make may be outdated. “Look up and gauge the applicability or palatability of outsourcing this stuff, and then review it each year because it changes.
Even with HIPAA or bank requirements, they are even now outsourcing. Are you reading the tea leaves from 10 years ago or from today? And ask yourself, what’s going to be politically acceptable ten years from now?”
When it comes to ceding control to a colocation provider, Carey suggests that you “ensure that your network team is in your discussions: First, the power discussion, second, the network discussion. You need to make sure that there is robust connectivity back to the colocation provider.”
Lim from Vantage Data Centers added, “Control is a relative term. Customers have access to our portals. Yes, you’re giving up some control over cooling and over management. We’re assuming that we can do it better [than customers].”
Next, Cruz asked the four panelists, “How do you feel about brokers?” Leonard and Brueggeman painted a largely positive picture of ViaWest’s and CyrusOne’s relationships with data center and telecom brokers.
“Some are incredibly helpful with getting customers further down the road. Some we love– about six of them,” Leonard explained. “Some of them are just there for 4% of the contract value, and 10% to 15% of the monthly value. Overall about half of ViaWest’s deals are done through brokers or partners. Clients need visibility about how much brokers are getting paid so that you know that you’re paying for it.”
Brueggeman added that CyrusOne “works closely with brokers. About 30% of our business is broker- or partner-led. [CyrusOne likes to] Make sure [it has a] direct line-of-sight communications with end customers. We need to make sure that we’re not only talking to the broker but also reaching out to the end customer.”
Although much of this panel discussion avoided going deep into any vertical market, a few points came up in passing; Cruz asked if colocation providers attract government buyers.
Keystone NAP’s Carey talked about how “local government in Pennsylvania is one of the slower segments to make decisions, and it’s heavily budget constrained. We can offer great value, but they get in the way of taking advantage of that value.”
Whether it is on the east coast or the west coast of the U.S., Carey and Lim from Vantage Data Centers are on the same page. “So many restrictions and contract terms, [providing colocation to government buyers] makes it very difficult. We don’t do any active outreach [with government buyers],” Lim added.
Coming from a larger colocation provider than Keystone NAP and Vantage Data Center, Brueggeman mentioned CyrusOne does “have state, county, and Federal service integrators one step away, but with the federal government, the much longer lead time makes it much more difficult.”
To wrap up this lively and information-packed panel discussion, Cruz asked: How do you spot bad colocation providers?
Lim jumped in first and asked half-jokingly, “Do you want the list?” I would not be surprised if Lim– or any of the other panelists– does keep a database, spreadsheet, or internal Wiki that catalogs and chronicles the shenanigans of the bad colo providers they have run up against.
However, then Lim more seriously added, “Physically visit. Talk to all of the teams. Look at the equipment itself. What is its age? Is it the end of life? Is there redundancy? What about uptime?”
Leonard’s tips: “Evaluate yourself: the procedures, models, and staffing. If you’re concerned about it, go with a broker to help you do the evaluation. There are not many bad operators for too long. They sort themselves out.”
Carey from Keystone NAP wrapped up the responses with similar– or at least consistent– recommendations and best practices to “physically see where the data center gets power from. Where is the closest substation? Are there multiple substations? How does the network come into the facility? Are there multiple facilities? SOPs (standard operating procedures) and MOPs (maintenance operation protocols) sort themselves out. You can’t change the location of the substation. Echoing some advice we heard earlier at today’s conference: if you can look up from the data center and see a cruise ship, don’t go there.”
Everyone comes to a conference like Data Center World with their own goals, plans, and challenges they are trying to address. This directly influences which sessions to attend when there are multiple concurrent breakout sessions.
From my perspective, for the kinds of revenue growth and differentiation challenges, I help our clients address, “Colocation Strategies – Questions Answered by the Top Tier Colocation Providers,” Liz Cruz and the four panelists from CyrusOne, Keystone NAP, Vantage Data Centers, and ViaWest just plain knocked it out the park.
The panel discussion defined colocation and covered close to a dozen well-timed topics including requirements around power, cooling, and density, public cloud, the pros and cons of colocation outsourcing, the downsides of building your own data center, market velocity, customer expectations, control, brokers, government buyers, and spotting bad colocation providers.
Did you attend this session at Data Center World? What did you feel was especially remarkable about the panelist's comments? What do you most agree or disagree with? Please sound off in the Comments section below.