Optimal performance

Data centre optimisation ultimately boils down to one thing – saving money. Steve Yellen of Aperture Technologies explains why organisations should optimise their data centres according to their needs, rather than the best technology available.

Money isn’t everything, the old cliché goes. But in data centres, money is everything. Economic cost is the biggest and most persistent concern that managers must face when running an efficient data centre. Data centres can consume resources, including electricity and land, at a startling rate, and the more centres a company builds, and the more technology that exists within them, the more ravenous they become.

Data centre optimisation is about running the best data centre you can for your money. The temptation is there to buy the fastest machines available, build the biggest data centre possible. After all, the logic goes, the more money invested in a data centre, the better it is, right? But data centre managers should be encouraged to think less about getting the best, and think more about what they need from their data centre. Therein lies the secret to good optimisation.

If a family man wanted a new car, he wouldn’t buy a Ferrari. Sure, a supercar would be fast, as well as hugely desirable, but it just isn’t appropriate to his needs. More likely he’d need a car with enough space for the kids and the strange assortment of seemingly random items that children accumulate. Similarly, the fastest, most expensive data centre is often completely unnecessary and would constitute a major drain on a company’s resources. A lumberyard, for example has very different needs from an international bank. The bank requires raw processing power in real time, and cannot afford to experience any downtime whatsoever. As a result, it can justify spending heavily to ensure near-perfect uptime. A lumberyard demands less of its data centre and might even be content with batch processing. The centre is unlikely to be running the intensive computations that the bank will, and any downtime, while undesirable, is not going to be catastrophic.

Key considerations
To effectively optimise a data centre there are three individual elements that need to be considered: power, cooling and space. It is important to make sure all three are under control and, importantly, appropriate to the requirements of the data centre. If they are not, then the data centre can spiral out of a manager’s control and start to become a major drain on the company’s finances.

The first step towards optimising a data centre is to optimise the power consumption. By making sure that a data centre is designed to work specifically to the needs of a company, and reducing unnecessary resource wastage, the data centre becomes more efficient, power costs are reduced, and so is the carbon footprint. At a time when the environment is a top concern, and wasteful companies risk being named and shamed, cutting the fuel bill can not only save face, but it can save money and help save the planet too.

Power problems can be exacerbated by the use of high density equipment, such as blade servers. Small and easy to set up, blades are enticing pieces of hardware but a data centre manager needs to ask himself whether he or she needs the machine, and whether the benefits outweigh the costs. The power demands of a full rack of blade servers, as well as the intense heat output they generate, can add significantly to a company’s financial burden.

Concerns about increasing power consumption in data centres have been steadily rising in the IT community, particularly since the blade server rose to popularity. Many data centres are reaching the point at which the power demands of the data centre exceed their available electrical capacity. These data centre managers are finding it hard to secure the extra power, often because there simply isn’t any more available to secure. This is the problem facing IT managers in urban environments such as London’s Canary Wharf; because of a major power shortage, they are being forced to find workarounds to the capacity problem, or invest in new builds. By optimising their data centres, organisations can get a clear overview of their available capacity and begin planning for future expansion if necessary.

The situation in Canary Wharf also highlights how the geographical location of a data centre can affect its operation. Data centres that operate in built up urban environments are often forced to do so in a very limited space. If this is the case, there is little room for physical expansion and close management of power consumption, cooling, and floor space use becomes essential. Inevitably, the limited space leads to increased density in the data centre, and cooling and power costs become significantly higher. Unless these elements are monitored, these costs can get out of hand and the business can lose money.

The second step in optimising a data centre is to review its cooling. Cooling is an essential, and expensive, process and modern equipment often comes with significant cooling demands. A recurring problem is that as more equipment appears in the data centre, and more heat is generated, cooling becomes less and less effective, and so the cost grows exponentially. By making sure the data centre is operating to a company’s needs rather than trying to use as much technology as possible for the money, this problem can be reduced, if not eliminated.

Optimising the space in a data centre can go a long way towards reducing costs. Spreading equipment out across the centre can prevent a lot of cooling and power consumption problems, and it’s a good optimisation tactic if the space is available. Unfortunately, the space isn’t always available. But when hotspots develop, there is a very real risk of outages, and creative use of space can be a more cost effective solution than cooling.

Building on success
Once an organisation has defined its optimal data centre, it can often cut costs by reproducing the same framework elsewhere. The fast food business is built on franchises, where there are systems for absolutely everything: from hiring and firing, to advertising design and how to assemble a burger. If organisations can document their data centres and processes, they can ensure standardisation, which in turn will cut the costs of ownership. Obviously, all data centres are unique, and what works for one data centre in a company might not work for another. But, by standardising the basic framework, the data centres become easier to track and manage, and so do the costs.

Ultimately, the only accurate way to run an efficient and optimised data centre is to know what equipment you have, the power consumption rate of it, the cooling requirements of it, and the relationship between all the elements. To deploy a data centre without having access to this information is reckless in the extreme. Those who go in blind risk becoming trapped in a never ending cycle of outages, hotspots and unidentifiable problems.

Power, cooling and space are not independent of each other. A change to the floor space can have a profound effect on the cost of cooling, and optimising power consumption can create a change in the available floor space. Optimisation is a delicate balancing act between these elements. Data centre management software, such as Aperture VISTA, is an essential tool, helping you to see the resource costs of the equipment and how they relate to each other. With this knowledge, data centre managers can not only optimise their data centres, but also have a good idea of the resources left to build on, thus preparing for the future.

Steve Yellen is Vice President of Product and Market Strategy at Aperture and brings over 20 years of extensive product marketing and technology experience to the role. He is responsible for product/market strategies, business development, product management, and product marketing.

Previously, Yellen served as Product Line Manager of Windows Products at DataViz. At DataViz, he brought the company's first Windows product to market, and with profit centre responsibilities, grew the product line to 40 percent of company revenue at a 37 percent annual growth rate. He has also held technical and management positions at McDonnell Douglas and IBM. Yellen has been an active voice within the technology thought-leadership realm throughout his career.