Virtualization Strategies for the CIOFor solid capacity management in a virtual infrastructure, the supply of host resources must always be greater than what your virtual machines (VMs) demand. But virtual infrastructures are dynamic, and a simple calculation can be complicated -- especially when it comes to storage resources. That’s where storage resource management comes in.
Requires Free Membership to View
When you register, my team of editors will also send you the latest expert resources covering all areas of server virtualization, such as platforms, architectures and strategies, server hardware, managing virtual environments, application issues and more.
Margie Semilof, Editorial Director
Capacity management calculations are relatively easy for processing and memory requirements, but getting useful supply-and-demand metrics from a storage layer is notoriously difficult. You need storage resource management tools to monitor I/O resources and avoid resource contention.
Storage capacity management: A whole new game
The goal of storage capacity management is to protect virtual machines with colocated storage
from storage oversubscription. You don’t want one VM with noisy storage use affecting others on the
same logical unit number.
Consider a situation where eight VMs are spread across four hosts. In a typical cluster, the processing of these hosts and VMs is separated, but their storage is not. To determine VM resource demand, most virtual platforms today look exclusively at processor and memory utilization metrics. The problem is that virtual workloads also exert pressure on their storage layer, so storage resource management is especially important.
Without proper storage resource management, one VM’s performance can affect others at the storage layer. Let’s say that seven of the eight hosts in a given scenario exert a nominal and equal demand for storage resources, with the eighth exerting significantly more. A cluster that is storage-aware can better distribute storage requests, potentially restricting them from the noisy machine while still servicing the others’ demands.
Storage resource management with vSphere 4.1
In vSphere 4.1, VMware has addressed storage resource management. Called Storage I/O Control
(SIOC), this storage resource management tool controls VM storage resource use. You can view
the process in the Resources tab under Properties for any virtual machine.
In previous vSphere versions, disk shares could be controlled only within a single host. The new storage resource management tool, however, enables disk shares to be controlled and managed among every host in a data store. Administrators can configure SIOC to set constraints on disk resource demands as a measure of disk I/O. These constraints are useful when I/O requests create capacity bottlenecks on a host.
Before setting shares in vSphere 4.1, enable SIOC within each data store, and set a threshold for what you consider resource congestion. This metric identifies the number of milliseconds of latency at which a storage connection is considered congested. If storage I/O resources reach that threshold, vCenter recognizes that a data store has experienced resource contention and will begin rationing storage based on configured shares. That means that VMs with higher shares will receive more attention than those with lower shares.
Like with processor and memory limitations, SIOC enables tighter control of storage I/O resources. Storage resource management is important because the constant changes to resource supply and demand in a virtual infrastructure make storage I/O resources particularly complex.
About the expert
Greg Shields is a partner and principal technologist with Concentrated Technology, an IT analysis
and strategic consulting firm. Follow him on Twitter @ConcentratdGreg.
This was first published in January 2011
Join the conversationComment
Share
Comments
Results
Contribute to the conversation