Irish-owned hosting provider Blacknight recently unveiled its new €1m data centre. Its CEO Michele Neylon explains why it chose to build its own facility, and the importance of pragmatism in making technology decisions.
As CEO of a technology company, how involved are you in technology decisions?
With my business partner Paul Kelly, who is CTO, we evaluate technical solutions on an ongoing basis. What generally happens with regard to technology decisions would be that our technical staff and some of our technical managers evaluate things, and present options to us. For example, the technical guys might think something is cool and shiny and wonderful, and the question I have to ask is: do we really need it? You can’t go haphazardly: ‘this is a good idea, let’s drop €50,000 on it’.
For us, the choice of technology we can make is critical. We’re providing services to businesses of all shapes and sizes, and for many of them, it’s mission-critical. So if we have an issue, they’re out; they’re offline. The number of businesses that are relying wholly or partially on their online presence to support a significant portion of their business is growing.
When businesses come to us, they do so because of the services we provide. So when we make a decision about the technology, it has to be something we can stand over. When you start putting stuff into a production environment, it’s real – it’s no longer some kind of academic thing. People are paying you money, there are service level agreements involved, there’s reputation.
Are you ever conflicted where you see a really cool technology but you have to veto it for the reasons you’ve just described?
There’s no conflict. The thing is, I have to be the voice of reason around some of these things. Sometimes I’ll come across something that I think is really cool … but if anything, I’m more likely to go the other way.
In the early days, our customers were technology early adopters. In some circles, beta is really cool. But if you’re hosting the backend for a big financial services company, beta is not cool, it gives cold shivers up people’s backs.
So it’s your role to understand the implications of decisions that get made?
You have to be aware and conscious of the consequences of making a change. If you’re charging somebody for a service, you have to be careful and make sure that it works.
Take your telephone or electricity provider: you want them to be reliable, more than anything else. Who gives a damn what technology they’re using? If you’re making the buying decision for a company, you have to think what will work that’s going to be reliable, solid and that’s not going to result in you being woken during the night to put out a fire.
Was this change in approach one that occurred over time?
There’s a very famous book in the technology sphere, Crossing the Chasm (Geoffrey A Moore’s book on marketing high technology services to mainstream customers). A lot of tech companies go through that, from early adopters all the way through. When we started out, a lot of our customers were geeks, we only offered Linux hosting, not Windows.
When we got past having five employees and we started winning the bigger contracts, that’s when it got really real. Probably for the first while, it was functioning at a level, but further down the line it became real, with real clients running real services.
Where is the line between the early adopter who’s comfortable with technology versus the mainstream business that just wants a service?
I don’t know where the exact line is, but I think it’s something you discover over time. If you don’t start drawing that line, and differentiating between the two, then you’re probably not going to retain those clients for very long because they’re not going to put up with it. Clients who in some cases are paying us thousands a year or thousands a month, require a stable operating environment and predictability.
We’ve been staffing up and bulking up parts of the management team. What you have to do is bring it to the next level, up the game, improve the quality of services and the way that we offer them. You evolve or die.
There’s so much acceptance of outsourcing as a business model, so why did Blacknight choose to build a data centre rather than just take more space in a third-party facility?
There’s a combination of multiple factors. At one level, there’s a cost factor: if you’re using the amount of data centre space that a company like us uses, you take space that never seems to cease to grow, and the cost just goes up and up. In terms of flexibility within the space, you’re constrained by what the third party will let you do, and what they have planned.
We’re seeing a significantly higher density of computing power per cubic metre, the hard drives are significantly faster, there’s more data moving around at much faster speeds. And the servers are pulling more power, so the power requirements in data centres have changed. But if you’re looking at an existing centre, they’ve built it so you can only draw a certain amount of power per cubic metre.
So when we were doing this, we designed it so that we could draw the kind of levels of power that we needed to draw. That’s another part of it: having the freedom to do it. Here, we’re building our own, we’ve got the control, and the flexibility. It was massive outlay initially but it will start to make sense once the data centre is at a certain capacity. In terms of things like efficiencies in power and cooling – you’ve got the term like free cooling, the ambient air temperature in Ireland is never particularly warm or cold. It’s nice and stable.
Were there any good-practice examples you were able to use for reference?
There wasn’t one specific example of a build we used as a template but we were conscious of what people were doing. One company came up with idea of a ‘data centre in a box’ concept: bloody expensive, by the way, but very cool. Instead of you building a facility, you would literally plug it in and it was done. It was the answer to everything.
Electricity bills have gone up massively, so we were very conscious of the power efficiency. In some of the older data centres, their power usage effectiveness (PUE) is 2, so for every kilowatt to run, it takes another kilowatt to cool. But you can get PUE that’s much closer to 1. That would be the ideal scenario, and we’re not there yet. We’ve pushed it down in our own data centre to about 1.2 or thereabouts and maybe we might be able to push it down even lower.
One concept we’ve not fully committed to would be to look at highly customised physical hardware. If you’ve built a data centre that is really efficient in terms of air flow, then why do you need to have fans on the servers? Maybe you can look at removing them, who knows? So we’re looking at a few options there. We’re also talking to the hardware vendors to see if they can help us to make that happen. For instance, how many units would we have to commit to, for them to build something custom for us?
Do you think there’s always a conflict between being close to the technology and then understanding the environment the business has to operate in? Do you have to approach the two with a different mindset, in some ways?
I think you have to get the balance right, and of course that’s the problem. Some people would probably be traumatised if they were exposed to all the aspects of running a business; the stress, the worry, cash flow … all those things are really hard and you have to deal with them. Being aware at some level and having some kind of appreciation for the business impact of a decision is important.
Some people might not understand that their choice, what they think is innocent, has an impact and a cost and that affects something. It doesn’t exist in isolation.
Trying to get people to understand that if you spend €15,000 on shiny hard drives that’s €15,000 that I can’t spend somewhere else, for example. That doesn’t mean we shouldn’t spend the €15,000 but you need to be able to rationalise it and justify it. A budget is meant to help you, not hinder you.
So are you more comfortable going over your planned spend on technology or do you still balk when it happens?
We would estimate how much we’re going to spend on hardware purchases per month or per year. It’s estimated and it’s not going to be accurate because you know damn well that it’s going to be linked intrinsically with volume of sales.
But no, I would balk at spending more. We would look at things like, say, timing: is there any point in buying this in this quarter or this fiscal year because it could mess our figures for the year? Might we be better postponing this expenditure by a few months? Whether it’s technology or something else, it’s expenditure. If you push that out, you’ve given yourself breathing room.
It’s an environmentally friendly data centre: can you briefly outline how you’ve achieved this, and do you think the ‘green’ issue is still a live one for your customers?
You will find some customers who do worry about these things, and we do get asked about our eco credentials. For some of them it’s a big thing, for others not so much. It’s a plus, but by building more efficient data centres, you automatically gain the efficiency and the green anyway.
A lot of what we would have done in terms of how we built this out was not so much to be green, it was more to keep the costs of running under control. So the two go hand in hand. But you have to be pragmatic.
Get your early bird tickets now!