BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
Each virtual machine in your data center uses a portion of the server's underlying computing resources. Optimal server consolidation depends on resource provisioning that meets the workload's needs. Adjusting resources manually, however, challenges even the most detail-oriented IT professionals. Automatic resource allocation alleviates mundane adjustment tasks, and the technologies have entered the market and quickly become more intelligent and accurate. But your IT staff still needs to monitor the way resources are used and balance workloads accordingly.
Ask yourself these three questions to consider the implication of optimal resource allocation and automated provisioning technologies in your own infrastructure.
What are the resource requirements for this VM?
Determining the resource needs of a VM helps you avoid performance bottlenecks caused by imbalanced allocation, but unfortunately, the requirements aren't always obvious. Tools such as Performance Monitor help admins in Windows Server environments identify major resource use, but application benchmarking also measures actual CPU, memory and network bandwidth use. Both methods allow you to more accurately allocate resources.
What happens if I over- or underallocate VM resources?
Resource allocation is a balancing act that can significantly affect both performance and IT budget. Overallocation wastes VM resources and limits the server's consolidation potential, while underallocation could halt VM performance altogether. Furthermore, VM resource requirements often change over time, which means that to avoid imbalance, you must practice regular monitoring and maintenance as well.
How can I automatically allocate resources?
As with many other facets of virtualization, you can now automate VM resource allocation. You can use both memory techniques such as memory ballooning and automation tools such as VMware Distributed Resource Scheduler to accomplish the task. Regardless of how you automate provisioning, be sure to accommodate for spikes in resource use.