One of the innovations recently announced by VMware is the optimised solution for Skype for Business.

As research and development department in Praim we decided to play a bit with it and make some quick tests to understand what the advantages of this technology are.

From an architectural point of view, the optimised VMware solution is not far from what other virtualization vendors have already proposed, it is about the client’s ability to create a point-to-point pointing connection without having to pass through the server.

Credits: VMware

The image better explains the concept: in the optimised case, once the connection is established, the audio and video call traffic passes directly from client to client without loading the network or the server. In this short article I propose to test experimentally and in a rudimentary way if the optimisation works and how effective it is.

In order to make measurements, I created a simple setup using two thin clients, a switch, a thin client with two network adapters to make measurements, and a couple of virtual machines.

A9014 thin clientwebcam Microsoft Webcam

Jabra audio peripheral device

Horizon View client 4.5

media engine (64-bit)

N9012 thin client

webcam Logitech C170

Horizon View client 4.5

media engine (64-bit)

VM 1Windows 10 LTSB 2016 64bit

4GB di RAM

Intel(R) Xeon(R) CPU E5-2630L @2GHz

VMware Horizon agent

Skype for Business (Microsoft Office Professional Plus 2013)

VMware Horizon Media Proxy 1.0.0186 (64bit)

VM2Windows 10 LTSB 2016 64bit


Intel(R) Xeon(R) CPU E5-2630L @2GHz

VMware Horizon agent

Skype for Business (Microsoft Office Professional Plus 2013)

VMware Horizon Media Proxy 1.0.0186 (64bit)

Misuratore di banda

2 network cards in Bridge mode

Ubuntu 14.04.05

bwm-ng used to make measurements

ServerHorizon View 7.2 (RTM)


Client-side optimisation test:

To check the actual operation of the installed packages when connected to the virtual machine, Optimized Mode appears (otherwise, failure to function is indicated by Fallback mode).

Another quick check is to go inside the Skype for Business configurator, and analyse the video devices. If VMware virtual webcam appears, this is used through real-time audio/video, otherwise, if the webcam is displayed with the device’s specific name, the optimisation is working properly.

Optimised modeNon optimised mode


At this point we just have to start a call and verify the behaviour in RTAV and through the new optimisation proposed by VMware. In this test we used Blast protocol, anyway I do not consider this very important because it is not significant to understand the possible benefit of optimisation.

The two graphs represent the band used by the bridge during a single audio call and a video call. The lines represent the call through two clients which take advantage of the optimisation, two clients without optimisation, and a mixed setup with just one optimised client.

The result is quite obvious, using the optimisation data that pass from clients to vm are very small. As expected, using a video call the result is even more evident.

Using just an optimised client, the situation is very similar to not optimising both clients.

Without having the presumption of having accurate analytical data, from these tests, it is easy to see how bandwidth optimisation is consistent and that the work done by VMware seems to immediately give its benefits.

The bandwidth peak measured during initialization, both in video call and audio only, remains to be seen.

However, it should be remembered that in all cases audio and video were of good quality, obviously this is also related to the setup which was in fact ideal, as during the tests all the relevant components: network, server,… were not loaded.

I wanted to verify the same operation even through a Linux client but the private version in my possession does not seem to be compatible with the released Horizon version of the release, support for the Linux version is given on Horizon view 7.2 that should be released shortly.


Praim thin clients