Unformatted text preview: Allocating inadequate VM memory may reduce
workload performance, while allocating too much
memory is likely to result in underutilization and
increased total cost of ownership (TCO). increasing TCO for our environment overall
We realized that we needed a method
for accurately determining the optimum
virtualization host server memory size, taking
into account factors such as workload and VM
hypervisor memory requirements, available The consequences of specifying too much 4 hardware platforms, memory costs, and our memory on virtualization hosts could include: consolidation goals. Memory Sizing for Server Virtualization White Paper Determining Virtualization Host
We created a step-by-step plan for determining optimum memory size for the
1. Analyze the maximum physical memory
actually consumed on business computing
servers running workloads that we plan to
migrate into VMs.
2. Select a candidate virtualization host
3. Analyze potential memory configurations on
the candidate server platform, including the
number of memory slots and the density and
cost of available memory modules.
4. Estimate VM hypervisor memory Physical Memory
We began our analysis by examining physical
memory consumption on production systems
representative of those that we planned to
virtualize. We also compared our results with
memory consumption data from other companies.
We used internally developed tools to collect
utilization data from more than 3,000 business
computing servers at Intel. The servers were requirements, as well as estimated savings running non-virtualized workloads on Microsoft from sharing memory pages between VMs. Windows*. They included file, e-mail, Web, backup, 5. For each potential memory configuration and database servers. identified in Step 3, use information from For each server, we estimated maximum memory Steps 1 and 4 to determine the consolidation consumption as follows: levels we can achieve.
6. Reconcile platfor...
View Full Document
This document was uploaded on 01/20/2014.
- Fall '11