Sizing private clouds gets patented  

Estimating the capacity requirements for a private cloud service can be just as difficult – and sometimes hit-and-miss – as it ever has been for on-premise infrastructures. And when an enterprise then wants to add additional shared or public services to the mix to build a hybrid cloud environment that estimation process often comes down to iterative trial and error.

So the US patent that has recently been awarded to Texas-based Gravitant could prove to be of some significance to such companies, for it concerns the development of a way to accurately predict the capacity requirements of private and hybrid cloud environments. The patent uses an analytics approach that allows customers to shorten planning cycles and base demand decisions on smart predictions.

“Using this patented technology, our customers have saved 30-35% in private cloud infrastructure costs by intelligently growing their capacity profile to meet demand while optimizing the overall IT Supply Chain”

“Using this patented technology, our customers have saved 30-35 percent in private cloud infrastructure costs by intelligently growing their capacity profile to meet demand while optimising the overall IT Supply Chain," said Dr. Ilyas Iyoob, Director of Advanced Analytics at Gravitant.

The analyst firm, IDC, estimates that approximately $22 billion will be spent on private cloud implementations each year by 2017, an investment profile that makes it important that enterprises size them correctly to meet expected demand. If the private clouds have insufficient capacity, they may not be able to meet the demands of rapidly increasing workloads and if they have too much capacity, it leads to wasted expense.

The problem is fairly complex because the private cloud runs many different applications each with a different virtual architecture and workload profile. This is further complicated by the fact that some applications can burst to the public cloud and others are not allowed to.

Traditional capacity planning techniques may work for simple application scenarios but are inadequate for predicting resource needs for applications with multilayer dependencies or those that include composite transactions, such as a single transaction comprising multiple sub-transactions). Determining a resource cost for such composite transactions is challenging for system designers. Guesswork and trial and error are usually used for private cloud sizing, which leads to under-utilised resources or bottlenecks as workload demands vary or increase.

Gravitant’s patent is aimed at providing a more accurate prediction method based on advanced analytics. Business demand is translated into IT application demand, which is translated into IT architecture layer demand.

The company uses advanced analytics to derive application and IT resource dependencies and then applies bottleneck identification algorithms to identify the capacity issues given a demand profile. The iterative nature of this algorithm then offers solution options, followed by a prediction of the next bottleneck. In this manner, the utilisation of IT resources can be more accurately predicted, thereby allowing the IT resources to be used more effectively without bottlenecks or under-utilising these resources.