Effects of virtualization and cloud computing on data center networks
4 
Table 2: Changing realities of VMs in x86 environments. 
Early deployments  2011 and beyond 
Consolidate two or three VMs per 
physical server 
10-50+ VMs per server 
Network bandwidth could be easily 
shared by a few VMs 
Higher network bandwidth requirements driven by: 
  Greater VM density per physical server  
  More powerful CPU-memory systems 
  VM mobility 
VM operates statically. Moving VM 
workloads is a rare event, such as when: 
-  Decommissioning servers 
-  Moving to higher performing server
Dynamic workload placement is common. Moving VM 
workloads occurs for events such as: 
-  Time-based VM creation for handling peak workloads  
-  Time-based workload shifting for optimizing power and 
performance 
-  Automated tools for facilitating workload placement 
VM workload remains inside a data 
center 
Disaster recovery requires VM workloads to move across 
physical locations (for example, to a disaster recovery data 
center) 
Moving workloads dynamically requires VMs to stay within a common VLAN in the same Layer 2 (L2) 
network. If you want to move a VM outside its L2 domain, you have to use manual processes such as 
assigning and updating the IP addresses for the failed-over services and updating DNS entries 
correctly. To provide maximum VM flexibility, many enterprises are evaluating ways to enlarge their 
L2 networks. 
New capabilities such as Virtual eXtensible LAN (VXLAN) and Network Virtualization using Generic 
Routing Encapsulation (NVGRE) logically extend an L2 network across L3 networks. However, even 
with this potential to move VMs across a L3 network, local traffic will still have higher performance 
and lower latency if it stays within a large L2 network.  
Client virtualization  
A specialized type of VM is the client virtualization technology such as virtual desktop infrastructure 
(VDI). VDI creates a client desktop as a VM. The VDI instance, however, is more than a simple VM. It 
includes the real-time compilation of the end user’s data, personal settings, and application settings 
with a core OS instance and a shared generic profile. You can either install the end-user applications 
locally as a fully packaged instance or stream them from outside the VM. Applications and user 
personality are injected into the core desktop VM, and a brokering mechanism manages connecting 
the end users to the VM (Figure 2). 










