Centre of power

Not so long ago datacentres were on the wane. Distributed computing helped bring about the decline of large data-processing sites that characterised an era of mainframe dominance.

But the tide is turning, driven by a focus on cost containment, the need to improve security and a persistent growth in hardware requirements to deliver software as a service to businesses.

The automation and consolidation of separate business processes ­ human resources, sales and marketing, production ­ previously led to a move away from the central mainframe into a distributed computing model based on industry standards and smaller departmental servers, says Ian Brown, senior analyst at Ovum.

“Designed a decade ago, these distributed systems were created for a different paradigm, one that did not have scale and availability at the forefront,” he says. “Today’s move away from the distributed system is leading to the creation of replacement datacentres.”

Gartner analyst Rakesh Kumar says the prominence of the datacentre to business strategy will increase during the next few years as IT service delivery moves further towards a more centralised model.

“The topology of the future mega datacentre will differ from that of the past,” he says. “Server rationalisation, hardware growth and cost containment are driving the consolidation of enterprise data-processing sites into larger datacentres. This will spur the growth of a more federated, hub-and-spoke datacentre topology for most large user organisations.”

For Royal Caribbean Cruises, an office relocation provided the impetus for a full-scale datacentre refresh to provide the business with the continuity and scalability required to support the UK headquarters and six offices across Europe.

IT manager James Mead says that although the central management facilities have created significant time-saving implications for his staff, reducing server footprint and power demands were equally important.

“The datacentre needed to be accommodated in a fairly restricted space, and had to operate efficiently to contribute to the company’s corporate social responsibility obligations,” he says.

Running costs and operational efficiencies are going to be significant factors in new-style datacentres. With many companies driven by the desire to reduce energy consumption, new server and storage optimisation technologies are being employed.

Ordnance Survey has already achieved a 38 per cent power saving in its datacentre by employing storage management technologies. The system automates the migration of large volumes of data from power-hungry disk storage technologies to removable media that draws no power.

But Dave Lipsey, information systems infrastructure manager, says difficulties in managing and backing up huge volumes of data, rather than power consumption, led the company to look for a better storage solution.

“We collect more than 40 terabytes of photographic data every flying season,” he says. “Backing up the information to disk was impractical, energy-inefficient and environmentally unsound. If we had continued to add new arrays in our internal datacentre at that rate, we would have had to update our power supply and heat removal systems in the near future. Not to mention the impact on space.”

Over the next few years, more companies will be looking to the datacentre to reduce operating costs and provide more help to drive the business. However, 58 per cent of respondents in a recent IBM survey on datacentre use said the biggest challenge for the chief information officer is the cost of operation in terms of power and people.

Roy Illsley, senior research analyst at Butler Group, says increasing demands for disaster recovery and high availability capabilities following recent terrorist attacks and natural disasters are adding to the burden of managing a datacentre. “This focus on cost demonstrates that most organisations are struggling to manage the datacentre efficiently,” he says.

Peter Sturrock, head of IT at data analytics provider Sumerian, says the desire to reduce carbon footprints will also drive the choice of technologies used in the future. “Technologies such as storage networks and virtualisation help us control our power consumption and costs, so any new technologies will always be under consideration,” he says.

The response from vendors has been to start marketing a wide range of technologies aimed at addressing these problems. Gm2 Logistics, a subsidiary of one of the largest paper merchants in the UK, has tackled the issue by concentrating on its datacentre footprint.

According to IT manager John Boyd, adopting the strategy has meant associated energy consumption being reduced by more than 30 per cent.

“Thanks to virtualisation we have eliminated more than a third of our hardware,” he says. “This means we are consuming significantly less power, which is great for the environment and financially beneficial.”

Server virtualisation is already proving to be effective at reducing costs and power requirements through consolidation and rationalisation.

But while many organisations are starting to look seriously at server and even storage virtualisation, most projects are still in the pilot phase and are not being widely used for mission-critical applications.

Complex virtualisation technologies, while able to provide the change the organisation seeks, require multi-faceted management skills. Although the deployment of virtual machines can be rapid, the expertise required by the IT department to manage the virtual world should not be underestimated.

Extracting more value from existing datacentre resources will be the challenge for IT professionals in coming years. According to Illsley, the focus will be on increasing the datacentre’s ability to meet the volume of business in an acceptable timeframe.

Such timeframes have to be met while reducing both costs and power consumption, through the implementation of an infrastructure that helps change the organisation to deliver the demanded added value.

Donald Crombie, information security manager at City of Edinburgh Council, says deploying a datacentre has led to improvements in IT service. “Our application and file servers have been rationalised and consolidated where possible and this has led to greater levels of performance and availability,” he says.

“The relocation of servers and their associated backup responsibilities from a variety of less secure locations has reduced the associated IT risk to the council. From a business perspective, it means that our IT services and applications are managed efficiently and are increasingly available where and when necessary.”

Datacentres are enjoying a comeback, but in a new and transformed guise. Capable of high capacity and efficiency, they are no longer the power-hungry beasts of the past, says Gartner’s Kumar.

“The topology of tomorrow’s datacentres will be dictated not only by the operational and technological efficiencies that drive lower service delivery costs using the smallest possible number of datacentres,” he says.

“It will also be dictated by the realities of the need to consider organisational, cultural and political issues, and a desire to take advantage of the IT investments made in recent decades.”

Next week: The second part of Computing’s definitive guide to datacentres looks at best practice in business

Watch our video on the key datacentre technologies www.computing.co.uk/tv

Datacentre technologies for the future

Datacentre in a box

Tomorrow’s datacentres are more likely to be delivered completely configured on the back of a lorry.

Customised shipping containers, stuffed full of server and storage capacity, with external water, power and networking connections, could soon be arriving at a car park near you. Sun Microsystems has already launched its offering – Project Blackbox – so watch out for similar products from the other major vendors.

Claiming to reduce the initial outlay in datacentre construction by 90 per cent and guaranteeing 20 per cent more power and cooling efficiency, these portable lights-out datacentres might not be such a crazy proposition.

Environmentally-friendly outsourced services

With the focus on power, cooling and maximisation of resources within local datacentres, there will be increased pressure to keep only the vital resources close at home and to outsource the less critical parts.

And if companies can operate in an environmentally-friendly manner, it also enhances their green credibility.

Data Íslandia is already leading the way, offering an international archival service from its Iceland facilities. Powered by geothermal energy and hydroelectricity, the service claims to be the greenest in the world.

If you do not need to keep your service in the corporate datacentre, expect to find another firm that can look after it more cheaply and environmentally cost-effectively.

Row and rack-cooling techniques

Borrowed from the telecoms environment, cooling requirements of the next-generation datacentre will be managed at row or rack level rather than taking the traditional room cooling approach.

The increasing rack densities, with their associated additional heat output, mean room-based cooling, which is dependent on mixing cool and warm air to prevent hot spots, no longer works effectively without consuming huge amounts of power. Integrating air conditioning units within rows or individual racks provides better cooling for higher density applications.

Lights-out datacentres

The concept of a datacentre that is self-managing and self-maintaining and can run itself with minimal human intervention has been around for a while.

Promised by vendors, the reality has not yet materialised. However, with continued advances in datacentre automation techniques, self-diagnosing and healing IT systems and improved management software with predictive tools will bring the concept a little closer.

It might not be feasible to achieve a truly lights-out datacentre, but we should certainly be able to turn the lights down low.

Intelligent datacentres

With power consumption big on the corporate agenda for cost and environmental reasons, vendors are looking for ways to reduce energy consumption.

Power-down storage solutions are going to be big news. By powering down the hard disks when not in use, intelligent systems claim to reduce energy consumption by 20 per cent.

Together with low-power server technologies designed to meet green computing initiatives, the modern datacentre is set to become less power hungry and more environmentally friendly.

Five companies to look out for

The Green Grid
A consortium of IT companies and professionals seeking to improve energy efficiency in datacentres. Promoting the adoption of energy-efficient standards, processes, measurements and technologies, The Green Grid aims to develop energy efficiency for datacentres. A useful resource for IT professionals tasked with datacentre operations, The Green Grid offers opportunities for sharing best practice as well as developing metrics for datacentre productivity, including computational efficiency.
www.thegreengrid.org

Smartbunker
In a remote ex-Nato nuclear bunker lies a 30,000 square feet, purpose-built datacentre with three-metre thick steel and tungsten reinforced walls: the perfect secure datacentre for firms looking to outsource datacentre activities. Smartbunker also claims to be the first and only UK datacentre with zero carbon emissions, using only wind-generated energy.
www.smartbunker.com

Optinuity
Offering a radical new approach to dealing with IT complexity, Optinuity brings autonomic policy management (APM) to the corporate datacentre. APM is about moving from a reactive mode of IT management to a proactive approach that requires no human intervention. Concentrating on making complex business systems self-healing and self-managing, Optinuity aims to solve business system problems before a chain of support activities is triggered.
www.optinuity.com

SprayCool
SprayCool technology has been developed by experts in thermal engineering and provides state-of-the-art cooling for computing and electronic systems. Using evaporative technology to maintain anything from single processors to complete rack-mounted systems at the correct temperature, SprayCool offers an efficient method of cooling. The innovative technology was first developed for military and government applications.
www.spraycool.com

Halcyon
Providing intelligent systems management and networking technologies, Halcyon has developed a portfolio of systems management tools to monitor and manage multiple midrange platforms using a centralised graphical console. Unlike many software companies, all Halcyon’s senior developers and commercial managers have worked as IT managers, so they understand exactly what the real world can be like.
www.halcyonsoftware.com

Best practice: Securely virtualising your datacentre workloads

Virtualising server workloads in your datacentre without implementing security best practice can result in
incidents that increase costs and reduce agility. Gartner analyst Brian Gammage outlines the following best practice when deploying server virtualisation:

Consider security when selecting virtualisation software
The host operating system should be thin and hardened – do not use a general-purpose operating system as the foundation for virtualisation. Try to avoid single point of failure dependencies on consoles or parent virtual machines (VMs). From the end of 2008, favour hypervisor-based systems in non-volatile storage, technologies that allow multiple versions of the operating system to run. In addition, take advantage of virtualisation-enabled hardware – use NX/XD-enabled systems and software that make use of such
capabilities.

VM vulnerability and configuration management
Make sure your application vendors support virtualised environments. In the absence of such support, plan for manual inspection and confirmation of configurations. Define your standards and lock down and configure VMs. Baseline your secure configurations and test against your policies over time. As a supplementary step, pressure vendors to support offline patching, software distribution and configuration management.

VM identity and access management
Tightly control audit, administrative and root access. Log all activities, link records to security information and event management (SIEM). Ensure security settings cannot be altered by operations. Define VM management roles and force vendors to fill the gap in VM identity and lifecycle management tools.

VM network access control
To enable network access control, do not share internet protocol addresses. Offline VM configuration management is a significant issue and you should press suppliers to provide support. For now, plan to manually update offline VMs and proactively shield VMs before they can connect.

VM intrusion prevention
Pressure network firewall vendors to support your virtual servers and protect offline VMs. Use a dedicated network card for VM management and set alerts on virtual tool use. Finally, ensure traffic between VMs that you do not want to occur really is not occurring.

VM business continuity and disaster recovery
Configure your systems to protect against denial of service and do not overlook disk space used by logs and queues. Externalise security and management policies wherever possible and plan for high availability when patching the virtualisation layer. Provide for inline security protection failure. Also avoid virtualisation architectures that introduce single points of failure, such as privileged consoles, service partitions and parent partitions.