Capacity Behavior Trends
Prism’s predictive analysis engine forecasts the capacity needs of applications running on a Nutanix cluster, giving the IT team the ability to proactively understand and plan their infrastructure needs. Prism’s advanced machine learning algorithms analyze and derive information from a broad spectrum of infrastructure telemetric to provide instant insights helping IT teams optimize their infrastructure budget at the same time ensuring business services never run out of capacity.
Prism continuously monitors usage of CPU, memory and storage across the cluster and accurately predicts when a cluster will run out of resources through what is called a capacity runway. This powerful capacity planning capability is powered by Nutanix’s patent-pending X-Fit algorithm. Unlike existing capacity planning tools that crudely extrapolate historical data patterns, X-Fit is application-aware and employs multiple predictive algorithms that continually compete with one another to determine the most precise forecast.
Existing capacity planning solutions use algorithms such as STL, ARIMA and Theta to forecast time series data. These algorithms do not adequately address the requirements and challenges that arise in modern day applications. Algorithms such as Theta, for example, don’t factor in seasonality aspect of resource usage and others don’t scale enough to use the vast amount of data points to build a time series.
Instead of using a single algorithm per application, X-FIT uses an ensemble of models and runs a tournament to select the best model for each environment. To use millions of data points and come up with an accurate prediction, X-FIT efficiently runs a distributed tournament for finding the best set of models for a given time series and optimally combines the forecasts from the best models.
With this capability, Prism accurately predicts when applications will run out of capacity enabling just in time provisioning of capacity.