The highly successful businessman Phil Crosby observed: “If anything is certain, it is that change is certain. The world we are planning for today will not exist in this form tomorrow.”
This quote has been referenced countless times and there is no doubt that it resonates truth. In the world of compute and data centers, change is often driven by customer demand. Businesses demand for data centers changes continually. The important question is whether data center operators are ready to adapt to that change?
When feet mattered most
It was not so long ago that the data center industry was thinking predominantly about real estate. Businesses would ask data center operators for a price to ‘house’ their compute. The data center operator would respond with a price per square foot or metre. For years data center operators were providing data center services priced relative to the physical space that was occupied by a given amount of compute.
When power was everything
Change is certain. More recently, businesses realised that the physical space that was occupied by a given amount of compute was less relevant than the power that their compute consumed. Data center operators had to adapt. Businesses began to ask data center operators for a price to ‘power’ their compute. The data center operator would respond with a price per rack or per kilowatt (kW). Power is a good metric of cost because it takes into account not only the physical cost of the data center, but the efficiency of a data center. Data center operators now tend to provide data center services priced relative to the power that is consumed by a given amount of compute.
The era of FLOPS
Things are changing again. In a world of continual pressure to do more with less, cost is the ultimate driver of decisions. Efficiency is more predictable than ever before. It is possible to calculate with a high degree of accuracy the compute resources required to perform a given job or workflow. As a result, businesses are now asking data center operators for a price to ‘run’ their compute. Driven largely by the cloud pricing model, the data center operator is now responding with a price per core per hour, which is ultimately used to satiate the customer’s rapidly growing need for the number of floating point operations per second ‘FLOPS’ (teraflops or petaflops) consumed by a given amount of compute.
A fitting misquote of Charles Darwin by Leon Megginson, both of whom would clearly have got on well with Phil Crosby, observed: “It is not the strongest of the species that survives, nor the most intelligent. It is the one that is most adaptable to change.”
Data centers are still physical buildings. The running cost of compute hardware that is ‘housed’ within them is still influenced significantly by ‘power’. Reliability and manufacturer competition has driven widespread commoditisation of compute hardware. In turn, this has driven predicability of the compute resources required for a given business. If availability of compute resources is expected to be 100%, the only key differentiator left is cost. If a financial services company requires a petaflop of compute resource to analyse inter day trading data, they will optimise for the lowest possible cost for that petaflop of compute resources. If a bioinformatician has a fixed budget for genome sequencing, they will focus on the maximum number of teraflops for their budget to sequence more.
As we move into the ‘Era of FLOPS’, some data center operators will struggle to change and adapt. At Verne Global, we thrive on being agile and understanding customer needs. We optimise for the lowest possible cost for data center solutions in the most efficient physical and operational environment. Whether our customers think about feet, power or now more likely FLOPS, our focus is always on what our customers really want: doing more for less.