vSphere 6 long-distance vSphere vMotion provides the ability to move workloads and to share and load balance IT assets and resources across physical locations, which has been a goal of IT departments since the introduction of virtualized servers. One of the main drivers for the continued adoption of virtualized servers is the control it gives businesses, along with the flexibility and agility that comes with decoupling the hardware and software from one another. This means that vSphere 6 functionality enables the development of new architectures that address specific customer use cases and can be adapted as required to specific customer and provider environments.

The solution proposed takes advantage of vSphere 6 long distance functionality and advanced network technologies to facilitate seamless virtual machine mobility. In Rainpole.com’s virtualized environment, the ability to dynamically and manually redistribute virtual machines to the new service provider’s location gives the opportunity for value-added services, operationally efficiency, and disaster avoidance.

The requirements for long-distance vSphere vMotion include that the maximum latency between the source and destination sites must be 150 ms round-trip time (RTT) or less, and that there are 250 Mbps of available bandwidth. For a complete list of requirements, see the VMware Knowledge Base article, Long Distance vMotion requirements in VMware vSphere 6.0 (2106949) at http://kb.vmware.com/kb/2106949

The virtual machine network must be a stretched Layer 2 extension, because the IP address of the guest operating system does not change during the vSphere vMotion operation. If the destination port group is not in the same Layer 2 address space as the source, network connectivity to the guest operating system will be lost. This means that stretched Layer 2 technology is a requirement. VMware does not recommend any specific stretched Layer 2 technology. Any technology that can present the same Layer 2 network to the vSphere hosts at each physical location will work because the physical network configuration is irrelevant to VMware ESXi™. Examples of valid technologies include VXLAN, VMware NSX® Layer 2 gateway services, Cisco OTV, and GIF/GRE tunnels. There is no defined maximum distance between source and destination networks that are supported by VMware, as long as the network meets the previously described requirements.

Architects must understand that long distance vSphere vMotion performance varies because it is constrained by the laws of physics. With one of the long distance vSphere vMotion requirements stating that the ESXi source and target hosts must be within 150 ms RTT from one another, consider the practical implications in defining Rainpole.com’s virtualized workload mobility use case with respect to how far the physical data centers can be located from one another. Therefore, the 150 ms RTT requirement must be translated into a physical distance between locations, and many variables exist in these calculations.

First, consider that the support requirement is a 150 ms RTT response time. From that it can be calculated that the distance between the physical locations is based on using 75 ms as the one-way support requirement. In a vacuum (a space that is devoid of matter), the speed of light travels at 299,792,458 m per second, and this figure can be used to convert to the time it takes to travel 1 km.

 

If the speed of light is equal to 299,792,458 m per second, the following calculations can be made.

 

 

 

From these calculations, it can be inferred that the speed of light takes 3.3 microseconds (μs) to travel 1 km (one-way). Therefore, it takes 6.6 μs to travel between the on-premises and VMware Cloud Provider Program data centers (RTT).

However, because the signal transmission medium on WAN networks is usually fiber optic, there is an interaction between the electrons bound to the atoms of the optic materials that impedes the signal. This increases the time it takes for the signal to travel 1 km. This interaction is known as the refractive index.

The following calculation is used when converting from a vacuum to an optical network.

 

The following calculation determines the distance that corresponds to 5 μs.

These calculations show that it takes approximately .005 μs to travel 1 km on a fiber optic network.

Therefore, light in a fiber optic network can travel approximately 1000 km in 5 ms.

Finally, consider that the VMware requirement is 150 ms RTT. That is, the time includes both the time to the remote data center and the return time. The round-trip distance that the speed of light can travel on a fiber optic network is double the one-way direction which can be reduced to 100 km per 1 ms.

 

The following calculation determines the RTT.

 

 

In conclusion, when considering RTT, light in a fiber optic network can travel 500 km in 5 ms, or 200 km (Rainpole.com’s target distance) in 2 ms.

It is critical to understand that there are a number of additional factors that can add additional latency across a network, and this increase in time must be considered. This added latency is unique to each environment and network provider. This latency is also dependent on factors such as signal degradation, the quality of the fiber or Layer 1 connection and also the different components that are between the source and destination endpoints. Therefore, VMware recommends that site interconnect options for the environment in question are evaluated by speaking to the service providers to further define these performance metrics.