TCPWave-DDI - Forecasting Charts Using Machine Learning

Introduction

In today's dynamic era, the allocation of resources has become a primary concern in network infrastructure monitoring. So, there is a need to monitor and predict the usage of the performance metrics such as CPU, disk, and memory of various appliances within the organization to avoid network performance issues, overprovisioning, and under-provisioning. Traditional forecasting methods are challenging to detect resource usage in real-time data. Hence the organizations look forward to a solution to assess their resource usage based on the historical pattern that helps them manage their IT infrastructure in a better way. This whitepaper provides insights into TCPWave's Forecasting Charts Using Machine Learing within the IPAM application.

TCPWave's Forecasting Charts Using Machine Learning

Network infrastructure monitoring with an automated forecasting system is an add-on for an organization to effectively manage network performance and business continuity. The following section explains the forecasting of IPAM, DNS, and DHCP Appliances performance metrics.

SARIMA Time Series Machine Learning Model

Seasonal Autoregressive Integrated Moving Average (SARIMA) is a time series Machine learning model used to forecast future data points using past data points. The model has two sets of integer parameters, denoted by (p,d,q) and (P, D, Q, S). The second set is the seasonal parameters.

The historical time series data is used to forecast the performance metrics of the appliances based on daily, weekly, and monthly frequencies.

Global Policy Configuration

To enable the prediction of the performance metrics in the TCPWave IPAM application in the chart format under the Performance Management >> Charts section, you must set the global option Enable ML Forecasting Charts to Yes by navigating to Administration >> Configuration Management >> Global Policy Management. By default, the global option is set to No. On setting the global option to Yes, you can view the ML forecasted data in the grey shade as shown:

Scheduler Management

To get the instant plots, you must execute the schedule jobs for the first time after setting the global option to Yes. Otherwise, wait for the scheduled job to be executed at their scheduled time.
Navigation: Administration >> Scheduler Management >> Scheduled Tasks

Conclusion

With TCPWave's forecasting charts, one can identify the trends and patterns using historical data. Using these trends, the network administrators can detect the outliers using Median Absolute Deviation statistical model that helps separate genuine insights from seasonal variations. It ensures enhanced security and reduced risks associated with excessive resource usage, safeguarding your organization's entire IT infrastructure, and is integral for meeting compliance needs. For a quick demo, contact the TCPWave Sales Team.