Workload Signature Display

Last Updated OnOctober 30, 2023

Feature Description

HEAL performs workload behavior correlation where the workload on the system at any point in time has a direct influence on the system behavior. Key system metrics are a direct function of the workload HEAL processes at any point in time. Consequently, the expected behavior of the system and even a component in the infrastructure are a function of the workload. During every baseline, a Machine Learning Engine (MLE) forms the workload clusters. Each workload cluster has a different transaction composition. Each workload cluster has a different impact on the amount of resources consumed across nodes.

HEAL showcases the application usage pattern in terms of workload cluster composition for a given time period. You can select the time period between 1 day, 1 week, 1, 2, or 3 months. HEAL displays the top three workload cluster compositions and the rest of the compositions under ‘Others’. ‘Others’ is all clusters together except the top 3. With every MLE baseline, the number of clusters may vary.

For Example:

Say, in an application, there are transaction occurrences of three different types. Say, T1, T2, and T3 are the types of transactions. MLE comes up with various clusters like C0, C1, etc. Say, C0 contains 50 transactions of type T1, 70 transactions of type T2, and 20 transactions of type T3.

Every minute, there are transaction occurrences of different types. MLE fits in the minimum of transactions mix to any of the clusters it creates during baseline. The percentage value for the cluster indicates that for a particular time interval, transactions occurring in the application belong to which cluster maximum.

Day wise selection