Keep your datacentre in good health

The datacentre is the heart of the organisation. The network provides the arteries and the veins that distribute vital information to the various parts of the corporate body, but it is the datacentre that keeps pumping it out. Keeping the heart in the best of health is critical to your long term survival, a difficult task with the stresses and additional burdens of the modern business environment.

The datacentre is the heart of the organisation. The network provides the arteries and the veins that distribute vital information to the various parts of the corporate body, but it is the datacentre that keeps pumping it out. Keeping the heart in the best of health is critical to your long term survival, a difficult task with the stresses and additional burdens of the modern business environment.

As data loads increase and place ever more strain on the heart, how can organisations ensure that they keep themselves in the best of health? Amid much talk of optimisation and virtualisation technologies, there are some much more basic questions to be asked about how you manage your datacentre. Are you going to build it and run it yourself, incurring potentially significant in-house cost and overheads at a time when budgets are under increasing presure? Will you outsource it wholesale, cutting your costs but potentially alienating your workforce and delivering a new set of challenges to do with managing a third-party relationship? Or perhaps something in-between - a managed service or an excursion into the brave new world of software as a service (SaaS)?

The reality is that whatever your preferred option is, there is no universal solution, but what is undeniable is that all organisations are facing similar challenges. According to one study by IDC, the cost of staffing and maintaining in-house datacentre operations came out as top concerns for those surveyed, at 62% and 45% respectively.

Another piece of research - Hewlett Packard's 2008 Data Center Transformation Survey - found that more than one-third of CIOs say that within two to five years, their datacentres will not be able to cope with growing demand for business services and applications. While the majority of respondents said that they spend less than 20% of their IT budget on maintaining datacentres, over a third expected this to increase over the next five years. This is a serious problem.

So some form of action is needed. What form that takes probably depends most on what the function of the datacentre is perceived to be. Is it an operational support structure used mostly for operational tasks such as transaction processing, or is it a centre for driving the creation of new processes and services to maximise the potential of the business, which takes it beyond the scope of basic transaction processing and storage? The goal of most datacentre transformation strategies is to be able to meet demands for increased datacentre capacity as well as reducing costs. There are multiple options for how to manage datacentre capacity and a number of key trends influencing the decisions made.

There are companies that still manage their own datacentres, especially in the hi-tech sector where they are providing services such as web-hosting for other firms. For example, telecoms firm Vialtus, formerly Pipex Business Services, runs five datacentres across the UK, including locations in London, Oxford and Leeds. Angus Peacey, head of product marketing at Vialtus, says that keeping tight control over datacentre management ensures that the firm can provide robust services to its clients.

To ensure that the datacentres operate at optimum capacity, Vialtus consulted with Pillar Data Systems to build network storage and managed service offerings around virtualisation, e-mail exchanges services and Oracle, a process which has resulted in standardisation on a Pillar-based infrastructure including Archive and D2D-based backup. As part of that process, the Pillar Axiom was deployed last quarter and is already at 60% utilisation.

Then there are those who have dabbled in outsourcing, but choose to bring datacentre operations back in-house because of changing circumstances. For example, telecoms services provider KPN has decided to in-source its Dutch datacentre services, which were outsourced to Atos Origin. "KPN is a very different company from what it was in 2001," says Marcel Smits, CFO of KPN, adding that bringing the datacentres back in-house enables the firm to engage with new business markets more effectively, boosting the bottom line.

Sometimes datacentre transformation is triggered by circumstances beyond your control. Northgate HR decided to abandon its Hemel Hempstead head office after it was partly destroyed by the explosion at Buncefield oil storage depot in 2005. The company, which provides human resources systems and services to some of the UK's largest employers, instructed local agents to search for a new building within a 10- to 15-mile radius, which meant rebuilding the datacentre.

Northgate HR's new facilities are in datacentres operated by IBM, which allows more customers to use the same server, meaning lower electricity output and lower costs. Typically, only between 5% and 10% of central processing units (CPUs) in a datacentre are used at any one time. Using IBM's P-Series, Northgate says the processing power is increased so CPUs can run at 80% capacity at any one time.

"Creating a next-generation datacentre does not mean building one from scratch - it is more a question of maintaining high standards in the interior of the buildings," says Wendy Logan, technical operations manager at Northgate.

"This includes conserving power, adjusting air conditioning, fire safety, doors and security. For many customers, modernisation also means updating to the latest technology, take for example the IBM P-Series, which means that we can service more customers on a single site.

"Normally, CPUs in datacentre hardware systems are not fully utilised, which incurs unnecessary costs and uses excess power. The reason this has to happen is because users have to allow for a certain amount of 'white space', which is only used in busy periods."

The datacentres are mirrored with duplicate facilities in London Docklands, and in West London. In the event of another disaster the size of Buncefield, customers' servers will remain running and could be back to normal service within hours, rather than days or weeks.

"Following the Buncefield explosion, our disaster recovery system has become known as a 'hot swap'. We operate two datacentres - one in East London and one to the West. Data is replicated from one to the other regularly, so in the event of one datacentre going down, we can swap all activity into the other within a few hours.

"Increasingly, our customers are also asking us for new datacentre measures to comply with Sarbanes Oxley requirements, including encryption technology for connectivity, and additional security around sensitive files. This has to be part of the datacentre," Logan says.

Organisations can opt for full-scale outsourcing of their datacentre operations, but, increasingly, there is a trend towards managed services, where certain aspects of the operation will be handed over to third parties.

According to research for COLT, half of UK businesses are already using some degree of third-party datacentre management today and 48% plan to increase this immediately. The research revealed that the main challenges UK businesses face in managing in-house datacentres include finding available space, security and managing the building and systems.

"The traditional IT model was to build and operate datacentres in-house, but CIOs are now increasingly leveraging third parties to meet their needs. The traditional IT model just is not receptive enough to the needs of the business. Line of business people cannot wait 18 months for processes to be delivered through the IT department. CIOs are looking to develop platforms that enable them to selectively deliver applications on a managed service basis," says Geoff Gilton, head of products for COLT Managed Services.

The current slide into recession is also adding to the third-party managed services shift. "One of the reasons that people look to third parties is to reduce the capital constraints of building a datacentre," says Gilton. "That is also one of the main drivers behind the SaaS movement with which CIOs are starting to engage."

The basic principle of SaaS makes perfect sense. If we want electricity in our homes, we do not dig up the garden, build a generator and produce our own power. Instead, we go to a company that provides electricity, buy as much or as little from that company as we want and then pay monthly or quarterly.

SaaS differs from wholesale outsourcing, says Zach Nelson, CEO of NetSuite. "The traditional application outsourcing model takes applications that were designed to be run on a customer's premises, and moves the burden of managing those applications to a third party. All the traditional outsourcing model achieves is shifting the cost associated with managing an application from internal employees to an external provider," he says.

The SaaS model does make you dependent on the provider's own datacentre being up and running, and there have been some embarrassing outages in the past, most notably Christmas 2005 when's datacentre went down in the back of a database problem. But then, say proponents of the SaaS model, your own server could go down on Monday morning with the same effect it is just that with the SaaS approach, you cannot just ring down to your own datacentre to complain.

The latest factor to influence datacentre management policies is the emergence of the Green agenda. "In the UK, focus on the Green Agenda has rapidly gained momentum," says Derek Wade, executive vice-president at Atos Origin. "Datacentres are believed to consume 20% of large corporations' energy requirements. Virtualisation, shared service models, utility computing, datacentre optimisation and waste energy re-channelling need to be employed to reduce the dependence on multiple equipment. Energy efficiency must be gained through combinations of all of these developments."

Datacentre infrastructure suppliers have realised that the "Green box" is an important one to tick off in any procurement and are adjusting their pitches accordingly. So, for example, Fujitsu has spent £44m on a 65,000-square-foot London-based facility that has a low carbon footprint as a result of using new IT-cooling and processing technologies, saving enough electricity to power as many as 6,000 homes every year, equivalent to saving 10,000 tonnes of carbon dioxide a year.

The Green agenda is also likely to increase the move towards handing over datacentre management to third parties. "For most large enterprises, running power-efficient datacentres is not a core competency," says Ian Brown of research house Ovum. "What better way to attack energy efficiency than by outsourcing the problem to an IT services supplier's shared datacentre? The leading datacentre infrastructure suppliers are all itching to sell their services."

So, in-house, out-of-house, somewhere in between? The approach taken to datacentre operation and capacity management will depend on your company's needs. It is up to you whether you go for homeopathic medication or traditional NHS pharmaceuticals the important thing is to keep the heart beating strongly.

Read more on Server hardware