Current Filter: >>>>>>

   

Most enterprise IT leaders are under pressure to deliver resources to the business faster and with the same level of service, or better. The challenge lies in ensuring that costs don't spiral out of control. The competition for internal IT is heating up with public clouds becoming a viable alternative for organisations, so cost competitiveness is paramount.

Such demands have made automated provisioning the go-to model, enabling businesses to capitalise on its ability to spin up new virtual machines (VMs) in a matter of minutes and inject substantially more speed into their operations. Yet automation per se isn't the fast track to infrastructure utopia. Incorrectly applied, automation can actually cause a host of problems with VMs roaming freely across the infrastructure and underperforming.

What's important here is how exactly the system’s automating infrastructure operations are controlled, and this is why there is so much focus on the role of analytics. Analytics have become essential in these complex operating environments with so many variables to determine how VMs should be configured and where they should be placed. The wrong decisions can lead to a variety of issues, from compliance and policy violations to serious performance issues and the inefficient use of resources.

Many current technologies claim to provide infrastructure control through analytics - however a lack of understanding as to how a solution really goes about its job can have major repercussions. One of the most prominent pitfalls is the use of rudimentary, simplistic, reactive balancing tools that claim to optimise infrastructure. These tools base their operation on short windows of utilisation data and, as a result, cannot properly manage the complexity of the environment.

Such tools have no concept as to how to optimise infrastructure, but rather operate by responding to issues that arise incrementally, moving or adjusting one VM at a time. This can throw infrastructure into a constant state of chaos, never getting to a calm and optimised state. Such solutions may act quickly but they will only kick in to respond to a specific problem as it emerges: in effect working as little more than a sticking plaster.

The result is excessive VM motioning, creating more volatility with workloads moving around again and again, and an infrastructure performing at just a fraction of its capability. Furthermore, the repercussions can go deeper than a lack of efficiency.

Constrained by their ability to access only the most recent data by offering a more limited perspective on the environment, reactive tools are not able to provide an entirely accurate summation of what is happening or what is really needed and can oversimplify the problem and remain blind to the real solution.

Determining the optimal approach to VM configuration and placement relies on in-depth information based upon a complete world model of the infrastructure environment. This is where predictive analytics are cementing their status as the only viable solution underpinning next generation infrastructure environments.

Applying predictive analytics that can take into account a myriad of technical, business and utilisation factors and constraints provides organisations with a roadmap to the optimised state. Leveraging historical utilisation patterns, personalities, profiles and application policies to predict future behaviour ensures that placements and configurations are designed to minimise risks, and it enables a plan that extracts the best possible use from the infrastructure.

With the right infrastructure view, organisations can confidently automate optimisation to free up stranded capacity for more workload, significantly increasing efficiency. At the same time, reactive VM volatility has been proven to be dramatically reduced, saving organisations from moving production workloads during active business cycles, in response to performance issues. Predictive control truly gives organisations the best outcome with reduced risk and maximum cost savings. NC