I have long had what some regard as an odd viewpoint on monitoring performance in desktop environments—which, when viewed from a traditional perspective, could be considered the case. To me, desktop monitoring covers all areas of performance monitoring, whether of physical desktops or of virtual devices delivered by way of a remoting protocol such as RDP, ICA, or PCoIP. It should be known by now that my personal view is that the only true metric is that of user perception. However, we all know this is a very difficult metric to measure, what with EUC performance being like beauty: existing in the eye of the beholder.
Now, I am by no means advocating that EUC monitoring is a dead duck. However, I am going to investigate what should be done to put in place a performance monitoring stack that can go a significant way toward fulfilling this requirement. Hopefully, it will ease the pain points in ongoing support.
First things first. End user computing (EUC) is not just virtual desktop infrastructure (VDI). It covers the gamut of devices, processes, and access methods needed to consume a desktop device. EUC is used for physical devices (i.e., on my desk or remotely in a data center); for virtual devices running on a hypervisor in a private data center (VDI); and for public, private, and hybrid cloud instances (DaaS). It also is used for presentation sessions managed via Citrix XenApp or Microsoft RDS sessions and application virtualization or layering technology, Mocrosoft’s App-V, VMware’s ThinApp, VMware’s Mirage and Unidesk’s Unidesk product. Each of these discrete products and touch points need monitoring.
Traditionally, monitoring has gathered information on the performance of your compute platform, your network, and your storage. Although these are valid metrics, they do not cover all the bases from a desktop perspective.
What are the application load times and the logon and logoff times? How quickly does YouTube load and run those funny videos of cats? What about my eBay page? How soon can I get from pressing the power-on button in the morning to actually doing productive work? How long do I have to wait for the machine to power off before I can leave at night? These metrics are more important to your users than the standard graphs that sysadmins use to show management that everything is working within limits. Your graphs may show no issues, but if your users are complaining, then guess what? You have issues.
Now that I have spent such a long time saying what I do not think is a good fit for desktop monitoring, let’s examine what is.
What Are the Requirements for a Desktop Monitoring Solution?
It depends on perspective. At minimum, it should deal with the following:
From the user’s perspective:
- Logon times
- Logoff times
- Application load times
- Browsing speeds
From the sysadmin’s perspective:
- The hardware that the desktop is residing on (virtual or physical)
- It must report on network latency
- It must report on network bandwidth
- It must report on storage latency
- It must report on storage bandwidth
From the business perspective:
- License compliance
From security’s perspective:
- Who is running what
- Who is accessing what
- When is what being accessed
- Who is installing/uninstalling what
Boot times, although important from a physical perspective, are not so critical to record in a virtual environment, where your desktop is already powered on and just awaiting a login.
What Products Meet These Objectives?
Truthfully, a single product will not be able to meet all the requirements.
VMware’s vROps for Horizon handles the infrastructure side very well, but completely ignores the client-side devices and the user-perspective viewpoint. The same goes for VMTurbo and its suite of products.
Liquidware Labs’ Stratusphere UX can provide analysis of desktops both virtual and physical, but not of the virtual desktops’ endpoints, and it does not provide insight into the underlying infrastructure.
Aternity’s APM suite does a very good job of monitoring both physical and virtual desktops. If you are a Citrix house or use MobileIron or Good Technology, you can monitor your endpoint devices, too. VMware has AirWatch, which controls endpoint and application access, but there appears to be no monitoring solution for it.
To sum this situation up, monitoring the end-to-end processes for EUC solutions is not a straightforward affair, as there is no product that will fulfill all of our requirements for end-to-end monitoring. Because we have no “one size fits all,” we need a mix-and-match approach to any end-to-end monitoring solution: for example, SCCM and Aternity for Citrix, or vROpS and Stratusphere UX for VMware View. But remember, just because it is easy to monitor your disk, CPU, and memory, this is not the be-all and end-all. Remember your user. Forget their needs at your peril. If I were your CTO and you came to me with a bunch of graphs that only monitored your infrastructure when you were arguing there was nothing wrong with your environment, I would not be amused. If I cannot watch my kitty videos at full-screen high definition, I really do not care that your graphs say there is nothing wrong.
Forewarned is forearmed.
Share this Article:
Latest posts by Tom Howarth (see all)
- That Was the Year That Was: 2016 - January 16, 2017
- Docker Has Been in an Acquisitive Mood Again, This Time Pulling in Infinit - January 9, 2017
- Acquisitive LANDESK Bought Out - January 5, 2017