Virtualisation – the catalyst for green data centres
Sean McCarry, UK and Ireland country manager, Novell explains how virtualisation technologies can enhance green efforts, but management is essential for real success
The last few years have seen organisations look much more closely at their energy consumption and in particular, the largest drain on energy resources, IT infrastructure. However, the recession has impacted the speed by which organisations have sought to develop more green IT and many data centres remain the energy gluttons, wasting costly – and limited – energy resources.
Adding to actual hardware costs is the increase in energy prices that has seen the cost to power data centres in Western Europe rise significantly, with IDC estimating the cost increasing 13 percent year on year in 2008 to reach €4.9 billion. For some low-end servers with an average three year lifespan it will cost more to power these machines than to actually acquire them. Consequently it’s clear that the data centre is a major consumer of power, not to mention a major contributor to any company’s energy bill.
But the pendulum may finally be swinging back as confidence returns. Industry commentators and analysts believe that green initiatives have started coming back onto the agenda because of the rise in costs for both IT and energy. As a result, CIOs and IT directors have turned their attention to the latest technologies for ways to reduce costs and build greener IT infrastructures. Attracting considerable attention is virtualisation.
A step in the right direction
Virtualisation consolidates the workloads of individual servers and runs them on a single, efficient server, requiring fewer physical servers and lowering electricity and cooling requirements. Many organisations have an abundance of home-grown or legacy applications running on individual, underutilised servers. Virtualisation allows enterprises to consolidate them on less expensive commodity hardware without having to rewrite old applications. For example, Sony Italia replaced 12 older physical servers with just two new ones running SUSE Linux Enterprise Server with Xen virtualisation, reducing its costs and freeing up valuable space in the data centre. The company avoided having to buy, maintain, power and cool ten new machines – and with 64-bit virtual servers running on the highly tuned SUSE Linux Enterprise Server platform, the new two-server cluster offers all the performance Sony Italia needs.
Why do many organisations have too much computing capacity in the first place? The answer is that historically IT has acquired a server for each application with a net result of creating huge server farms where each server has a utilisation rate of only about 20 percent. With virtualisation, IT can substantially increase the compute-capacity utilisation by using the same hardware to run multiple applications independently. This means that utilisation rates can rise towards 80-90 percent, allowing IT to reduce the number of servers. For example, Essent, the Dutch energy company, undertook a major data centre consolidation project, virtualising a large number of servers to reduce costs and energy consumption. As a result, Essent believes that the solution saved them about £2m for the data centre consolidation project alone.
If it sounds too good to be true…
While the case for virtualisation has been made industry-wide, its benefits come hand in hand with more complex server management issues. This benefit and cost offset has inhibited the full utilisation of virtualisation. Today, each virtual machine is managed as if it were physical. In order to achieve the lower power consumption and cooling costs virtualisation promises, organisations must also incorporate automated virtualisation management.
Implementing virtualisation alone is like assembling an orchestra without a conductor. You can have the best violinists, trumpet players and harpists in the world, but without an experienced conductor, chaos will ensue. Similarly, with virtualisation, a single point of management keeps all systems working together and supporting the business. Ironically, while attempting to reduce physical server sprawl, organisations can inadvertently create virtual server sprawl. Doing so likely means a host of unanticipated capacity and resource allocation issues. Understanding how to manage and allocate effectively is vital to optimising the new arsenal of virtual machines. This is where automation tools come in to play.
IT regularly ‘brings down’ a server for updating or servicing. While an inconvenience, workers have come to work around this procedure. With virtualisation housing a number of tasks and applications on one server, many aspects of the business are going to be affected by these downtimes. And on the off chance there is a server failure, there is a risk that a large portion of the business is completely ‘off-line’.
This scenario is a nightmare from an organisation’s perspective but especially so for businesses that are prediction orientated such as manufacturers since any downtime in the supply process caused by downed servers can lead to lost output, unsatisfied customers or lost sales. It would take only one of these ‘offline’ instances for executives to reverse their position on virtualisation in their data centres to avoid these disastrous situations.
Automated management alleviates the heavy manual process of moving files and applications. Not only does this help avoid an ‘offline’ situation, it enables effective server maintenance without risk. And if a server, either physical or virtual, fails there is automatic, rapid deployment of services for business continuity. Organisations can plan scheduled maintenance with confidence and reduced hassle, while keeping the business running smoothly. All the while ensuring cooling and power consumption savings and reducing the impact of the data centre on the environment.
Putting it into practice
Pernod Ricard Pacific, a wine and spirits company, is a great example of this in practice. The company was moving to new premises in Australia with a much small data centre and energy capacity. By migrating existing physical servers to virtual machines on SUSE Linux Enterprise with Xen, and provisioning new virtual servers instead of buying new hardware, the company eliminated or avoided buying a total of 50 servers. Instead of the projected 58kW, the company is operating all existing and new services inside the 32kW limit imposed by the new data centre and still has room for growth in terms of floor space, power and cooling.
In addition to reducing hardware acquisition and maintenance costs, Pernod Ricard Pacific manages the physical and virtual infrastructure from a single solution which automatically provisions new environments based on the available resources. Moreover, the increased utilisation of the hardware resources means that the company can accomplish more useful work within a smaller power envelope reducing operational costs and cutting the carbon footprint. It estimates that with Xen virtualisation on SUSE Linux Enterprise Server, it is saving 625 tonnes of CO2 emissions equivalent to planting 2,250 trees.
Virtualisation has been on a crusade the last few years but with the emphasis returning to both cost reduction, IT efficiency and greening of the data centre, its importance will only rise further in the coming years. While virtualisation can be a tremendous boost to an enterprise’s productivity and environmental policy, without the proper automated management tools in place, it will threaten the very benefits sought in the first place.
Virtualisation’s promises of reducing sever sprawl, heating and cooling costs, and power consumption are enticing, but they cannot be attained without effective management in place. With orchestrated management tools that automate critical data centre processes, organisations can make virtualisation a central component in their green IT strategy.
