VM Storage I/O Visibility
Ensuring storage I/O capacity is critical for proper VM performance. VKernel offers solutions that can show which datastores are facing high I/O utilization or identify VMs generating large storage I/O loads in shared storage environments.
vOPS 4.5 Overview
Storage I/O Contention Affects VM Performance
Monitoring storage I/O capacity is critical for proper VM performance. Gaining visibility into this infrastructure area can be challenging as storage-level monitoring tools show disk perfomance but have difficulty in tracking which VMs caused storage I/O spikes. Furthermore, it is important to identify what specific actions within the VMs caused disk traffic increases in order to resolve issues. Observing storage I/O at the VM, host and datastore level is critical to maintaining a high-performing virtual environment.
Storage I/O Performance Issues Can Occur in Several Infrastructure Areas
The underlying storage infrastructure used to support a virtualized environment has its own set of capacity and performance bottlenecks. These bottlenecks can be located in:
- the host HBA
- the fabric or network
- the storage array controller
- the storage array back end
- or the disks themselves
Storage I/O challenges can be identified broadly by throughput, latency or IOPS issues. The challenge is to gain sufficient visibility into storage I/O and the VMs generating it to rapidly identify and resolve VM performance issues stemming from storage. This visibility can show whether a virtualized server administrator can resolve the problems or if more intensive action by the storage team is needed.
Resolving VM Performance Issues Stemming from Storage I/O Bottlenecks
Similar to the challenges presented by general VM performance issues, detecting and correctly diagnosing storage I/O problems requires more than just real time monitoring tools. VKernel's Capacity Analytics EngineTM collects and analyzes over 20 metrics for each VM across resource pools, hosts and clusters on a standardized time interval which includes disk and network metrics. For a typical 100 VM environment, that is over 2000 metrics.
vOPS Performance Analyzer then uses the results from the Capacity Analytics Engine to identify and resolve current VM performance problems originating in storage areas for the virtualized environment. Performance Analyzer will screen for issues related to:
- Disk latency
- Queue latency
- Disk throughput
- Memory swapping which may affect storage performance
Visibility into the virtualized infrastructure is provided at the datastore, host, and VM level to further aid in troubleshooting. Additionally, various alarms based on storage I/O metrics can be set to alert virtualization adminstrators of potential issues. Learn more here about vOPS Performance Analyzer.
