Optimising Utilisation Forecasting with AI and Machine Learning
What IT team wouldn’t like to have a crystal ball that could predict the IT future, letting them fix application and infrastructure performance problems before they arise? Well, the current shortage of crystal balls makes the union of artificial intelligence (AI), machine learning (ML), and utilisation forecasting the next best thing for anticipating and avoiding issues that threaten the overall health and performance of all IT infrastructure components. The significance of AI has not been lost to organisations in the United Kingdom, with 43 per cent of them believing that AI will play a big role in their operations.
Utilisation forecasting is a technique that applies machine learning algorithms to produce daily usage forecasts for all utilisation volumes across CPUs, physical and virtual servers, disks, storage, bandwidth, and other network elements, enabling networking teams to manage resources proactively. This technique helps IT engineers and network admins prevent downtime caused by over-utilisation.
The AI/ML driving forecasting solution produces intelligent and reliable reports by taking advantage of the current availability of ample historic records and high-performance computing algorithms. Without AI/ML, utilisation forecasting relies on reactive monitoring. You set predefined thresholds for given metrics such as uptime, resource utilisation, network bandwidth, and hardware metrics like fan speed and device temperature. When a threshold is exceeded, an alert is issued. However, that reactive approach will not detect the anomalies that happen below that threshold and create other, indirect issues. Moreover, it will not tell you when you will need to upgrade your infrastructure based on current trends.
To forecast utilisation proactively, you need accurate algorithms that can analyze usage patterns and to detect anomalies—without false positives—in daily usage trends. That’s how you predict usage in the future. Let us take a look at a simple use case.
With proactive, AI/ML-driven utilisation forecasting, you can find a minor increase in your office bandwidth usage during the World Series, the FIFA World Cup, and other sporting events. That anomalous usage can be detected even if you have a huge amount of unused internet bandwidth. Similarly, proactive utilisation forecasting lets you know when to upgrade your infrastructure based on new recruitment and attrition rates.
A closer look at the predictive technologies reveals the fundamental difference between proactive and reactive forecasting. Without AI and ML, utilisation forecasting uses linear regression models to extrapolate and provide prediction based on existing data. This method involves no consideration of newly allocated memory or anomalies in utilisation patterns. Also, pattern recognition is a foreign concept. Although useful, linear regression models do not give IT admins complete visibility.
AI/ML-driven utilisation forecasting, on the other hand, uses the Seasonal and Trend decomposition using Loess (STL) method. STL lets you study the propagation and degradation of memory as well as analyze pattern matching whereby periodic changes in the metric configuration will be automatically adjusted. Bottom line, STL dramatically improves accuracy thanks to those dynamic, automated adjustments. And if any new memory is allocated, or if memory size is increased or decreased for the device, the prediction will change accordingly. This option was not possible with linear regression.
Beyond forecasting, ML can be used to improve anomaly detection. Here, adaptive thresholds for different metrics are established using ML and analysis of historic data will reveal any anomalies and trigger appropriate alerts. Other application and infrastructure monitoring functions will also be improved when enhanced with AI and ML technologies. Sometime in the not-too-distant future, AI/ML-driven forecasting and monitoring will rival the predictive powers of the fabled crystal ball.
by Rebecca D'Souza, Product Consultant, ManageEngine