DIGIT: What drives the Scottish data centre industry?
Kenny Lowe (Head of Emerging Technology at Brightsolid): At a micro-level, it varies from region to region and city to city. If you take Aberdeen for example, a major driver there is cost-saving after the oil price fell, but that’s also coupled with an additional desire to drive more value from their existing data. At a more global level it is more data analytics, machine learning, IoT, things like that.
But these are only really things that become available when things come together under one virtual roof. When data is disparate and fragmented you cannot drive as much value out of it. So actually one of the key drivers in data centre growth is getting data together under this one virtual roof so we can do analytics and machine learning, etcetera.
If you take Brightsolid as an example, we opened a new Scottish data centre in Aberdeen, and just a year or so on from that we have over thirty customers there. Aberdeen City and Aberdeenshire Council have wholly moved into our data centres and we have some great case studies there about how the experience has been transformational for them, and how that translates through to their end user experience – which is where it matters.
Chris Shields (Regional Sales Director for Scotland at Pulsant): There is a big drive to offer more than co-location as companies move towards a ‘hybrid IT’ model – where a company will have some of their infrastructure on-site and some of their infrastructure in a data centre, some of their infrastructure in a private cloud and some of their infrastructure in a public cloud.
Now customers are looking much more to take infrastructure as a service and manage that infrastructure through public cloud. Part of that is driven in ‘economies of scale’, Pulsant looks after around 40,000 servers, so we have a massive economy that can scale right into the market of Scotland.
Also what is driving public cloud is that we’re beginning to use our data centres a lot more – they are the stopping points for public cloud. One of the big driving points of that is Microsoft, they are about to release a product that takes their public cloud offering Azure, and condense that into a discreet piece of public infrastructure that we can drop into our data centres. In essence we could start offering Azure in our data centres in Scotland. What we saw with small companies is that they didn’t have the scale, but companies like ourselves can now provide infrastructure on their behalf.
Graeme Gordon (CEO at IFB & Chairman ScotlandIS): As a Scottish data centre operator IFB is a part of the 2008 ScotlandIS Naturally Cool initiative, developed to promote Scotland as a great place to build and develop commercial data centres. The main benefits being stable politics, geography and climate. A stable and predictable, relatively cool climate means less costs spent on power to cool your data centre – which is one of the biggest operational costs.
Naturally Cool is composed of two parts: the first was a Scottish Government wish to attract mega data centre operators to Scotland – such as Facebook, Google or Apple. The second was to encourage home-grown Scottish data centre operators to build and accommodate the need at that point to host data close to end-users and edge content providers near to data user and creators – like IFB and our clients in oil and gas.
DIGIT: How have customer requirements changed over the last 3-5 years? What have been the key trends?
Kenny Lowe: I would say that needs have changed significantly. If we go back to just three years ago, our approach would have been predominantly infrastructure-up. We would look at how much CPU, how much RAM, how much storage a customer needed, and just provision that for them. Today we take an application-down approach, so understanding where an application should live, for example in the hyper-scaled clouds like Azure or in our data centres, or on-premises, because each of these is a valid option for specific application workloads. I think we’re doing a lot more work with customers now than we were doing in the past.
Chris Shields: I think it’s become more complex, it’s not an ‘either or’. The need is becoming more complex to provide a hybrid IT environment where some things will always be on site or co-location, but also trying to make use of scaling infrastructure as a service.
Pulsant is seeing about a 10% growth in demand for co-location (where the customer will bring their own equipment), and about a 40% growth in demand for cloud services (where we provide a computing platform for clients and the clients use that). The majority of our growth is in cloud. I think the global co-location market has grown at about 8% give or take, and cloud growth is at about 40%, in line with the European market. We typically look at the UK market because we have data centres throughout the UK, and the growth is the same across our facilities in Scotland and down South.
A server needs power, needs to be cooled and needs to be secure. A data centre concentrates on making all of those components as resilient as possible, so there are diesel generators outside, batteries to flatten the load and make it more predictable, as well as cooling on a massive scale. Our electricity bill is massive, so what we try to do is drive that down with something called PUE (power usage effectiveness) – how much electricity we need to supply into the environment to power the computer – the lower that number is, the more efficient you are.
Fergus Kennedy (Head of Compliance and Information Systems at Pulsant): Environmental impact and management is part of our concern. We have a social responsibility to ensure that we are as clean as we can be. ‘Environmental aspect’ is the term – this is the potential you have through operations to impact the environment. As Chris points out, our largest aspect is electricity consumption, so therefore our main measures are around responsibly sourcing that power, and ensuring that we use that power in the most effective way. This centres around making sure that the infrastructure costs for facilitating the delivery of that service to a customer is as efficient as possible, ultimately also saving the customer money as well.
DIGIT: How have data centres evolved in recent years to consumer demands and infrastructure changes?
Chris Shields: There are two parts – there’s the data centre aspect and the access aspect. On the data centre aspect, Pulsant has invested over £3,000,000 to make sure we’ve done our bit. On the access part, if you look at businesses in general they are well-served. Businesses have a large choice if you are in or around cities. Some cities have more choice than others, but if you take Edinburgh and Glasgow and the big towns in-between such as Livingstone, they are well-served.
Graeme Gordon: Connectivity looks after itself mainly – that’s why we have invested in dark fibre not only in-country but also in the Trans-North Sea, where we can connect and peer with the rest of the world without heading south. Not all our eggs are in Manchester or London. So we are geared up and have been for some time to securely and resiliently push and pull huge amounts of data from our own Scottish data centres and also to and from partner data centres in and out of the UK.
Kenny Lowe: That’s a good question, because in most cases they haven’t evolved. I read a report from a service provider in the States which triumphantly proclaimed that AWS (Amazon Web Services) is a rip-off, and that you could save up to two-thirds by running dedicated servers in their own data centres, instead of virtual machines from AWS. That completely misses the point of why people buy into a service like Azure and AWS. They offer ancillary and development services, and things like machine-learning and serverless computing – hundreds upon hundreds of services that you just can’t get anywhere else.
While data centres like ourselves were deploying virtualisation solutions – bringing customers services and infrastructure to run virtual machines – at the side hyper-scaled data centres were evolving at pace. So things like Amazon Web Services, Microsoft Azure and Google Cloud Platform created a breadth and depth of services that just don’t exist anywhere else. But what Microsoft realised is that cloud shouldn’t just be a location which is their data centres, it needs to be a model you can apply to any location. So there’s now a trend of bringing these hyper-scaled services back to the ‘edge’, back to the regional data centres, and delivering value from them there. And that’s an exciting project to be involved in.
DIGIT: How do you envisage the future of the market?
Graeme Gordon: There has been a lot of Scottish data centre build activity since 2008, and there continues to be. This is great for buyers as there is a lot of empty space, as cloud has disrupted this space massively. There still is and will continue to be the need for data centres close to clients for operational and compliance reasons, but smarter networks and smarter application from the likes of Microsoft and Google means you simply don’t need as much space for servers.
If we want to attract a mega data centre to Scotland we have all the tools, skills and people, a good climate, stable economy and politics, great networks and potentially through hydro-electrical power the ability to provide very cheap, green power with long term predictable costs.
Kenny Lowe: I think in terms of physical rack-space needed in data centres, that’s going to continue to decrease. If you think that we used to deploy thirty to forty physical servers in a rack, now in that same rack we can run two- or three-thousand virtual machines, and each of those virtual machines can be more powerful than those thirty to forty physical servers ever were. So the amount of value that we can derive from physical real estate in a data centre will just continue to exponentially increase, which means that the need for really large data centres is on the decrease. That’s from a regional data centre perspective.
For hyperscale providers like Amazon and Azure, that’s where the large-scale data centres will exist. I guess just right now we’re starting to see the impact of things like Internet of Things (IoT), and all of the data and network traffic that gets generated as a result of that, and regional data centres can and will act as data aggregators and first responders in an analytics path there. In my view if you don’t have a network edge strategy, you don’t have a cloud strategy. So actually, regional service providers will become this critical-edge component to the hyperscale providers over the next few years.
Chris Shields: I think we’ll continue to see a move to more cloud. More clients have their business applications inherently fitting the cloud, starting from scratch and never having bought their own hardware. They will then get to a point where they will want to bring their cloud services physically closer to home for performance reasons, and also to have better governance wrapped around it.
Fergus Kennedy: That is supported by the relationship with Microsoft around the deployment of cloud accessibility with Azure Stack. A jump purely to public cloud is a significant step for a lot of businesses. But what we will be deploying is a localised, understood and compartmentalised public cloud capability, such that a customer is able to get themselves involved and evaluate how they can use that effectively.
The appetite is growing – the whole move in the market to the hybrid IT model demonstrates the desire to ultimately use the right tool for the job in each element of business. That might be public cloud for high resource consumption short-term processes, and then there might be regulatory or other requirements that mean they need to keep closer tabs on some elements of their business or data assets, such that they need that spread of hybrid IT. The localised data centre house cloud, whether that be a private cloud within a rack or an Azure-specific node, gives them the ability to take a collection of services that builds up the most appropriate hybrid model for their requirements.