vmware_horizon_suite logoEver since VMware started selling View as a desktop solution, we’ve had these discussions with colleagues and customers. You probably are familiar with the topic:

Many companies have more than one location, but they all want to manage their IT infrastructure as if it was one. So, if we can throw in a big data-line between those locations with low latency, highly resilient, including rapid fail-over and so on.. can we create one big View desktop pool for all our users spanning all our locations?

The answer is: No you cannot!

“But the bandwidth is 10 Gbps and the latency is very low!”

It doesn’t matter, you still cannot.

“Why??”

Because of latency. Latency is the key. A WAN connection isn’t as fast as a LAN connection (no really, it isn’t), and you simply need that speed to keep all the services up to date. For instance, an infrastructure with two sites will have two or more connection servers, at least one per site. These servers need to keep track of the status of VM’s as well as allocation events. To do this, they use the Java Messaging System or JMS. Now, JMS is a very fast system, but latency can kill it.

Imagine, a user logs into Connection Server 1 at site 1 and is allocated Desktop 1 in Floating Pool 1. At that same time, another user logs on to Connection Server 2 at site 2, selecting the same pool, Floating Pool 1. If this notification is not as close to instantaneous as possible, the second user could be incorrectly allocated Desktop 1, causing an error message for one of the users when there is a clash of login. Latency, Jitter and other WAN induced conditions can cause this transfer of data to not reach the other Connection Servers in a timely manor, resulting in many different issues.

This example was shamelessly taken from Simon Long’s weblog. If you want to read on and get the in depth explanation of how it all works and what you should do, check out this superb article by him. Simon does a terrific job in explaining why you can’t have that stretched cluster.