Leveraging data centre infrastructures

Anthony Foy, MD of Interxion, explained to FST why using a third-party data centre can pay dividends

FST. From the end-user perspective, how have data centre technology changed in the last 20 years and what advantages do today’s data centres have over ancestors?

AF. One of the things that has certainly changed is, in the way that data centres are being designed with an inherent ability to be upgraded. So they are much more flexible.

Historically, the data centre was built to deliver a certain amount of power and cooling capacity – In fact this was hard-wired into the design of the centre itself. However, the technologies we have today mean we have the flexibility to change the distribution of cooling or power throughout the data centre. For example say we had 10 megawatts of power going into a building we could effectively move that power to different parts of the building and focus it where we want to. This means that as applications or hardware deployments change over time you don’t lose the investment you’ve made in the data centre, and that you’re data centre is flexible enough to grow and evolve in line with requirements.

Another big change we’ve seen is a dramatic rise in the amount of power used per square meter in the last five years. The change has been from about 70 watts per square meter average usage to nearing about two kilowatts per square meter. And as we look at the forecast from our customers, over the next two to three years we’ll be pushing towards the three to four kilowatts per square meter mark, so there’s been a dramatic and ongoing growth in power and corresponding cooling utilisation.

FST. And has that increase in power been due to the number of mission critical applications increasing or is it more a function of the technology becoming more powerful?

AF. Both actually. There is a broad and deep IT refresh that is happening right now in that more companies are moving mission critical applications to external data centres. As a part of that refresh the most effective technology available tends to be either pizza or blade servers that stack up fairly densely in a rack. Blade servers are smaller, high performance servers that have a smaller footprint then conventional servers. The downside is that the more you put in a rack, the more power they utilise. So there are two thoughts to deploying these servers. One is to spread them across a large data centre, the other is to deploy them in a dense configuration in water cooled or enclosed cabinets with forced air to cool those environments, this allows you to stack up to 20/25 kilowatts The net result of the deployment of blade servers you’re getting effectively five to seven times more computing power into a standard rack (which would normally operate at between three to five kilowatts).

Because of the ability to concentrate higher levels of power with in a single cabinet customers are loading their data centre with a lot more power. This allows the companies to take advantage of the original investment that was made .

With reference to software, data centres are increasingly being used to house mission critical applications such as HR, CRM or billing . Rather than having multiple applications or databases running in different geographies or different places in the company, many companies opt to have a single instance of that application running inside of the data centre that multiple users from multiple geographies are able to access. The applications themselves haven’t changed radically, but the way they are distributed has changed.

FST. Does leveraging shared infrastructures help drive down costs? What’s the advantage of using independent data centres that are handling more than one company’s data?

AF. This is the really value proposition of outsourcing your data centre requirements to an organisation such as Interxion. It’s effectively a time-sharing model where we invest in the CAPEX up front in order to leverage our purchasing power and build a much larger data centre than any one company would need to use. We can then pass those savings onto the customer, and we have a core-competency in running the data centre.

A shared infrastructure also means that you can drive down the top three operational costs associated with operating a data centre once it is built. These costs are the people, the power and the real estate. Out of all these costs the cost of people is growing faster because finding good personnel is becoming harder as many more organisations are leveraging data centres -  highly skilled personnel are hard to come by.

Further more as footprints of equipment are shrinking and less and less data centre space is needed, many organisations will be very hard pressed to use more than 150 to 250 square meters of data centre space.  

At this size below this it is completely uneconomic to build a your own  data centre because many of the components needed to build a robust data centre environments would have been designed  and priced to meet the demands for large deployments .  For example - you can’t buy a generator that will just support a centre that utilises 250 or 500 kilowatts – you’re going to have buy a generator that provides a minimum of a megawatt. So then half or more of your capacity will have gone to waste.

By sharing the infrastructure, the generators, the coolers, the batteries, the people, the resources to manage and operate the data centre, and leveraging the cost of purchasing the power, you can makes big savings. We’re not talking about sharing computer resources – each different company will have a discrete secure area within the data centre, but they’re all sharing the same infrastructure.

 
FST. Financial services firms have traditionally treated data centres as a core competency, how can they benefit from managed services from third parties?

AF. For a long period of time, possibly the last decade or so, there was a real competitive and comparative advantage that companies were able to realise by deploying IT applications and infrastructure. But now the technology has developed to a point where it’s now more standardised, replicable, common approaches to the type of equipment and applications they’re going to be running. So there is less and less competitive advantage in managing and maintaining your IT infrastructure.

The competitive advantage is, and will remain, in the way you use those assets – so the way you use the data and the information, the way you design your databases, the way you use business intelligence, and the way you interact with your customers – all of that will always be a core-competency that a third-party wouldn’t be able to step into. But the base infrastructure of the data centre – keeping the power on and keeping it cool – is no longer a competitive advantage, it’s just a chore like having you car cleaned. Somebody has to do it, and if it’s done properly your business stays up and running, and if it’s not it doesn’t.

There’s no advantage in developing a core-competency in keeping systems on it’s much more cost-effective to go outside of your organisation.

Further more some third part data centre suppliers such as Interxion, arte carrier-independent, we are not bound to any particular network or carrier. So companies locating at our data centre have a multitude of choices of carrier network solutions. This allows them to realise the most competitive prices for bandwidth on the market. In many cases we’ve seen customers who have been able to realise savings in their network solutions that more than cover the entire cost of their outsourced data centres requirements

Anthony Foy joined Interxion as group managing director in July 2001. He is responsible for the overall strategic development of the company’s sales, marketing and business development activities across Europe.  During his time at Interxion, he has been instrumental in the turn around of the company from a point when it was losing €10 million per month to a position where it is structurally cash flow positive and profitable and has IT sector beating internal rate of returns.

Foy now plans to drive the double-digit revenue growth in 2007 and continue to grow the corporate customer base, introducing more new value added services in response to customer demand and enhance customer satisfaction to beyond the current 97 percent.

Prior to joining Interxion, he was SVP and General Manager international of Broadbase Software (now Kana Software), where he built the company’s international operations, sales and marketing infrastructure across Europe, Asia and South America and was instrumental in growing the company from start up to IPO and revenues of more than US$100 million.