The datacentre of the future is shaped by a number of forces including commoditisation, virtualisation, integration and innovation. A group of industry experts sat down to discuss the key factors shaping the evolution of the datacentre of the future and its impact on the channel.
Jennifer O’Brien (JO): What are some of the trends (and challenges) in the datacentre space?
Les Davey (LD): There is pressure coming from a number of different directions. One of them is the advent of cloud computing, which is driving a tremendous amount of activity in the datacentre space. We have got three different clouds that we’re standing up, a local T3 datacentre cloud, a global cloud and a Microsoft Azure based cloud. A tremendous amount of activity from the partner space as well and the intensity of the computing technology that is going into these datacentres is driving a real efficiency requirement and I guess finally the sustainability, the green environmental side of things is becoming very, very prevalent. There are so many elements and pressures coming into the datacentre environment because of the massive growth that we’re seeing and, of course, because we’re in Australia data sovereignty is becoming a real issue as well. People don’t want their data going offshore and people like us fortunately can support these organisations, particularly federal and state government and financial institutions who want their data local.
Gordon Makryllos (GM): Energy efficiency is a big issue. The cost of electricity is up 30 per cent, so energy efficiency is the key issue. The other one is high densities. Australia does lead in terms of blade server adoption and so there is much more high density in Australia than we’re seeing in Asia. Virtualisation, blades the whole density thing is driving new dynamics and we’re seeing now racks with a ton of equipment in it, so that has changed the whole structure from the very physical level right through to the energy efficiency. The increasing high density, the energy efficiency, the cost of power and customers wanting to maintain flexibility because they’re not quite sure where this whole cloud thing is going to go are some top challenges.
Maree Lowe (ML): A challenge is the actual lack of datacentre space in Australia, which is appearing in every state and at the government level. Apart from looking for actual datacentre space out there I think the challenge is for us to stop thinking about space, and to realize that it’s not about the physical footprint anymore. The smart way of building a datacentre is no longer about the size of the physical footprint.
Guy Carlisle (GC): I’d add one more factor to the planning when you’re looking at a datacentre and that is flexibility. You don’t want to build a datacentre and outlay all your costs at the start of the project. You want to be able to modularise and add or subtract as you’re going along these days.
JO: What are some of the customer pain points?
Peter Prowse (PP): Australia and the 30 per cent power increases is a major customer pain point. The key challenge for most customers is how they resolve that problem. They’ve already got significant investments embedded in their existing datacentres. They don’t generally have the $300 million lying around to go and build a new green datacentre, so how do they actually make those incremental improvements to drive down the cost of power and efficiencies? One of their biggest challenge and questions is, ‘What is the first thing I do to drive down power?’ And while high density computing solves space problems it doesn’t solve the power problem. Virtualisation doesn’t solve power problems in and of itself.
LD: I saw a study that said power in the next three years is going to go up 46 per cent, which is a significant end cost.
Ronnie Altit (RA): There is an upside to the power issue — it’s now getting on the roadmap of the executives. Previously IT was all about throwing service and buying more storage and throwing more equipment at it. Now it’s ‘What are the implications from a space perspective? What are the implications from a cost perspective’ because they’re recognising that IT — whereas historically the datacentre was usually the domain of the corporate services department — it’s now rapidly becoming the domain of the IT department. The IT department has to concern themselves with power. Historically that has not been their real strong point. With the looming threat of regulatory requirements and reporting for certain organisations, they need to start reporting on how they are becoming more and more efficient and that datacentre being at the linchpin of the usage of electricity. The upside is that now the executives are starting to understand the implications of what IT is actually going to require, what IT is going to drive and that is starting to add further fuel to the whole 'Do I put something in the cloud, is it my own cloud, is it someone else’s cloud?'
Richard Walsh (RW): Some of the information that is requested when speaking about datacentres is what size footprints, what power do these agencies want to use. It’s very hard to plan for it in contract terms. People need to start to look at ways to consolidate, use shared services to reduce their footprints.
Aaron Steppat (AS): There are a couple of challenges around reporting. How do I do the reporting of what I’m actually consuming? Are the systems and datacentres of today in a place to be able to do that? Typically we haven’t seen any action based on the reports and information that we have, so another large customer challenge is, ‘what do I do with the information that I now have?’ How do I go about gaining those efficiencies or driving down costs or making my staff more efficient to do more with less?
Joel Dawson (JD): When thinking about the ‘ideal datacentre’ there are so many elements to it from the power, the cooling, the density, the virtualisation. There are all these infrastructure elements, but then on top of that you’ve got the whole issue of, ‘How do I manage all this? When we increase virtualisation, then you start getting virtualisation sprawl. There are so many different elements that the IT managers of today have to get their head around and so things like automation, process, those are going to play a key part, as well as the whole issue of fixing the infrastructure and getting the infrastructure optimised. There is going to be that whole element of, ‘How do I manage and maintain and monitor this?’ And you can’t manage what you can’t measure.
GM: Measuring is a key factor. How do we measure at the rack level and at the device level? You have to figure out what is your carbon footprint — all of the big customers are asking for that now because it is part of their audit requirement. A lot of our opportunities around any existing datacentre is around measuring what you have — then there are the green opportunities. But a lot of people have to figure out what their heritage issues are and that is about measuring what you have today. A big opportunity and the other issue is around alarms. Hot spots are increasing. We find many of our customers do a terrible job of managing the environment and the alarms. Alarms are going off and they’re not paying attention because they’re not seeing it.
Cam Wayland (CW): If I was the CIO and somebody said my prices or my costs are going to go up 46 per cent over a period of time that’s going to change the business case. How do you guys that are talking to the customer say, ‘We can’t tell you what it’s going to cost to run in x years’ time?’ So how do they build a business case when you can’t give any certainty around that?
RW: That is why you get away from the facilities conversation. You don’t talk about the datacentre per se. You talk about the service that you’re providing.
LD: Looking at the smaller organisations, the question in my mind is how much will cloud computing resolve that for people? At the end of day, cloud computing is about being elastic. It’s got to be scalable and it’s got to be pay-per-use. You’ll be able to 'get on' and 'get off' as the market commoditises and then you will be able to have that portability and the access to service that as a small client you wouldn’t normally be able to get, so to some degree I think cloud could address part of it.
Sean Murphy (SM): For a small business, the messaging and archiving and backup and, in particular, network redundancy are above the mindset and the budget of a smaller business by and large. They would never build it themselves and when it’s an all-or-nothing option at some of the classier datacentres it becomes quite intimidating and it puts the cost up. No one is arguing it’s not a good thing, but it’s just taking them on a jump. It’s a journey to the cloud for the SMB. They’re not ready. They’ve got weird little associations between things. They haven’t actually finished getting into even VMWare virtualisation 101, let alone those dynamic HA and resource scheduling and getting everything else together and that is part of the journey. Their budgets are small.
LD: There is no doubt that cloud is bringing high quality datacentre capability down lower and lower into the market and making it more accessible, bringing high availability, robust environments, services and business practices into their organisations better than they could ever achieve by themselves or with the traditional model.
JD: I think the ideal datacentre for an organisation — and the promise of cloud — almost is that I don’t have one. From my perspective, the ideal datacentre is I actually don’t have one.
GM: We need to be careful because we shouldn’t say cloud is everything that isn’t on the customer site. The best research we’ve been able to get a hold of says within the next five years it’s only going to be five per cent of the IT spent will be on cloud services. Yes, it’s growing fast, but it’s off a very, very small base — there is still a lot of heritage.
I think there is going to be a lot of work before it meets a lot of the customer’s needs to move everything to cloud, whether they are buying a whole service or have local collocation. There is going to be a whole mixture there, so we shouldn’t assume that everything is going to rush to the cloud. We’ve got to be careful and not get caught up in all this hype because I think it is going to take a lot longer than a lot of people think and it’s going to be driven more by not necessarily people wanting to get rid of their IT burden, but because they want to dictate what applications they want. In the Australian market, a lot of people are rushing to provide hosting in collocation facilities because I think that’s the bigger demand right now than cloud.
Glen Hastings (GH): I’m concerned about the supply issue. As someone who’s in the business of sitting in the middle of all this stuff, sitting in the middle of the vendors and the customers and having investments in datacentre space for a considerable amount of time and making all that stuff work, that’s a concern. I’ve got customers today who’ve got racks of their own infrastructure in our datacentre connected by our network and managed by us, which when their refresh cycle comes around they say, ‘I no longer need four racks, I need one rack of someone else’s stuff’, and that’s a worry.
PP: I’m a little cloud weary and I think everyone else is as well. I’ve heard a lot of conjecture about people saying, ‘I can move everything to the cloud and that will solve all my problems’. Do you really expect cloud providers to be able to absorb 46 per cent power increase over the next three years and keep their costs the same? To me cloud is not going to solve customers' fundamental problems around high density, high power, carbon neutrality. For me the question is whether you are responsible for your own datacentre with your own organisation or whether you’re looking at moving that responsibility to somebody else — you still need to be looking at efficiency.
JO: Where else can partners help companies get cost savings.
GM: One of the biggest savings is the way cooling is done. The traditional mode of cooling is inefficient now with high densities. That can only scale to about three or four kilowatts if you’re lucky. Most racks, most designs that we’re doing nowadays is 7, 15, 30 kilowatts per rack, so by doing more energy efficient cooling and using hot-aisle containment, you can contain the heat in a smaller area and it much more energy efficient and cooler. By going into any traditional datacentre we can offer 50 per cent savings in most occasions just by doing cooling smarter.
It’s about keeping ahead and planning for it and we can do that by the way you set out your aisles. In the future if you don’t contain the heat, you’ve got the option. We’ve got the option to take out a rack and put in an additional in-row cooling device, so a lot of customers we deal with we’re designing it for when we go 30 kilowatts even though they’re not making an investment upfront. It’s about how you lay out the racks.
LD: The bottom line is if you’re using a modern innovative datacentre of which there are several available in Australia they’ll be running a closed loop cooling system. They’ll be using 90 per cent less water and less power and they’ll probably be using recycled water where they’re using water. And corporations can’t always design and deliver that sort of technology and they’ll be making other decisions on UPS technology, you know diesel versus lead batteries or some sort of rotary device and they’ll be constantly looking for refinements. But at the end of the day paying lower costs and then through the competitive nature of the market that is driven back as a new price point to the market, so they are getting a better service. They’re getting it reported properly. They’re getting more efficient services at a better price and they’re staying in front of a market against their own competitors because they’re paying the right amount for the best service they can get. And that’s not something that’s easy to build if you’re building a one-off datacentre for yourself.
GM: What’s interesting is the larger your organisation, like the big banks the less visibility the datacentre manager has on his energy bill actually. It goes to the building — it is part of the infrastructure. That is changing.
JO: Are companies going the green path because of the cost savings?
LD: I think there are a number of issues. The first issue is reputational. It’s really important for organisations to maintain their reputation as a credible organisation in their industry and then clearly there is an efficiency side of that and those things both spawn into their competitive ability to execute in the market.
PP: I think that we actually, as an industry, made a fairly fundamental mistake hooking onto the whole green IT thing in the first place because it is discrediting, in a way, the actual real benefits of power savings and cooling, which is dollars to the bottom line. And usually we didn’t come out and say they’re going to be carbon neutral, so they could look good for their readers. They worked out if they can drive down the cost of their operation that’s dollars to their bottom line. Green IT is another phrase like cloud computing, which is quite hackneyed and it’s around what is the cost to the organisation, what is the benefit to their bottom line of becoming more efficient. That is the marketing message I think needs to get out there a lot more.
GC: It’s all about return on investment. Companies will invest a certain amount to save a certain amount. That is why green is happening. It’s the cherry on top if you like or the icing on the cake. It’s purely an ROI calculation. Do I or don’t I and how much do I save by doing it?
JO: In addition to power and cooling, what are other areas of opportunity for partners?
GM: A verticalised approach in terms of design and development is a huge channel opportunity. There are 60,000 computer rooms out there. I think our big one is measurement. It’s tough to measure the environmental and power use, then once they start measuring it it’s how do we then make the next incremental improvement and then it’s taking a modular approach and having a structured architecture.
The biggest challenge is dealing with the heritage stuff and the fact it wasn’t architected. It grew organically and haphazardly. It t was more uniquely designed for that environment. It was uniquely designed for that floor, that building and that creates all sorts of challenges. APC’s value proposition is about having it modular, scalable and being able to plug and play. The biggest trend that we’re seeing is not having unique engineering at each site.
There are a host of challenges when developing or upgrading a datacentre. APC’s Gordon Makryllos highlighted a few key components of the datacentre of the future:
- Highly flexible with room neutral configurations so it can be implemented in any office room, eliminating the need for a dedicated computer room with raised floors and lets the datacentre be installed anywhere.
- Agnostic at the rack level so that it is not locked into any server or vendor architecture.
- In order to meet power and data demands of the future, the ideal centre would be designed from the outset with the flexibility to grow the server and power capacity in line with the business needs.
- With increasing demand for high density computing, the datacentre would be readily scalable to deal with high density requirements as people virtualise and take on new technologies and IT demands.
- Ensure the datacentre is cloud ready to provide flexibility to users. Datacentres to support cloud services must be able to expand in response to user requirements and ensure organisations only pay for what they use. The ideal datacentre will have infrastructure that can grow with the size of the cloud — as users grow, the infrastructure grows.
- With energy costs continuing to increase rapidly, the ideal datacentre would have to be energy efficient. The best, and often simplest, way to achieve this is through the latest cooling technology – in row cooling combined with precision cooling. With precision cooling, you only cool heat generating areas, rather than the whole room, creating huge power efficiencies.
- Incorporate infrastructure management software into the datacentre to provide continuous energy management and monitoring. This also includes an element of physical security, so that access to racks and devices is monitored and alarmed.