memory-sizing-for-server-virtualization

We expect to periodically reassess memory options

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Memory Assumptions When estimating total server memory requirements, we need to consider the memory required by the VM hypervisor in addition to any other overhead requirements. On the other hand, there will be savings due to memory sharing across VMs because we expect to run multiple copies of the same OS and applications in different VMs on the same server. For simplicity, we assumed that the additional memory required for the hypervisor would be offset by the hypervisor’s ability to share common OS and application pages across VMs. Achievable Consolidation Levels For the purpose of this memory sizing exercise, we calculated the consolidation levels that we could achieve on our hosting servers 7 White Paper Memory Sizing for Server Virtualization 32 GB Hosts 24 GB Hosts 16 GB Hosts 9 Top Quartile 7 4 10 environments, other factors independent of memory considerations can impose constraints on the consolidation levels. These factors include performance and risk management considerations. 21 Second Quartile based solely on memory constraints. We recognize that in production 16 Because we assumed no net impact of the hypervisor on overall memory requirements, the calculations were simple. We based 37 Third Quartile them on: 27 18 • The average of maximum memory consumption within each of 55 the four quartiles in our sample of more than 3,000 servers. 41 Bottom Quartile 27 0 10 20 • The three candidate target memory configurations: 16 GB, 30 40 50 60 Potential Number of Consolidated Servers 24 GB, and 32 GB. We divided each of our candidate memory configurations by each Figure 3. Potential consolidation levels by quartile, based solely on memory constraints. We calculated average consolidation ratios for each quartile by maximum memory consumption. of the quartile maximum memory utilizations to obtain a range of consolidation scenarios, as shown in Figure 3. For example, to determine how many of the servers in the third quartile we could consolidate into one 16 GB virtualization host, we divided 16 GB by 0.86 MB, the average maximum memory consumption for servers in 32 GB Hosts 24 GB Hosts 1...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online