AIOps should have Predictive Analytics

Predictive analytics has been the unique feature of ZIF. Machine learning algorithms used in the Predictive analytics module of ZIF have been granted a patent in the USA. The module uses unsupervised machine learning algorithms to predict a potential failure or performance degradation in devices or applications.

Once the performance data of applications (transaction time details) or devices (CPU, Memory, Disk, Network) has been ingested into ZIF using the universal connector, the algorithms start to learn the pattern of the data and look for potential anomalies that can cause failures or performance degradations. If issues are predicted by ZIF the same are created as opportunity cards in the platform for operation teams to take action.

The prediction dashboard has 5 swimlanes where the cards are populated. Based on the estimated time to impact the opportunity cards gets generated in wither warning or critical swimlane (if the estimated time to impact is less than 60 mins, the cards are generated in warning swimlane and if they are greater or equate to 60 minutes, then they are generated in critical swimlane)

If engineers have acted on the predictions made by ZIF, then the cards are updated to processed swimlane and if the engineers have missed taking an action, then they become lost cards.

Clicking on each of the cards, the details of the predictions can be viewed by the operations team. These details will identify the root cause of the issue that is prone to happen with the time and date when the issue would happen.

The operations team resolves the issues predicted by ZIF manually by logging into the application or device or can trigger automation bots to resolve the issues.