Key Concepts in Capacity Analytics > Optimal VM and vDisk Placement > Seasonality-based Trend Analysis

Seasonality-based Trend Analysis

Seasonality-based trends take into account the resource utilization patterns observed at different times of the day, week, or month according to your business requirements. This type of analysis is essential because applications have varying resource demand at different times. For example, in typical data centers, some VMs might experience high resource demands during the day whereas other VMs experience high resource demands during the night. The trend analysis is performed by using at least four to six weeks of the time series CPU or memory utilization data to generate the optimal VM sizing and VM placement recommendations for a given set of workloads (VMs and physical hosts) and VM hosts.

In the following figure, it is observed that five out of 10 VMs experience peak resource utilization during the day whereas the other five VMs experience peak resource utilization during the night. SHO considers this trend of resource utilization for each of the VMs and recommends five physical hosts for optimally hosting the 10 VMs according to their utilization pattern.

Apart from the historical trend analysis, SHO considers the user-defined constraints in the optimization scenarios for defining the VM placement plan. Constraints include specifying the headroom, whether VMs must be together or apart, and whether particular groups must be considered as exclusive. For the purpose of optimization analysis, all physical hosts selected in a scenario are converted to VMs.


© 2011-2012 Hewlett-Packard Development Company, L.P.