Beginner: Have Tera1 clients always been so slow?

[blackcopy]

Hi!

I've just started evaluating View 6 in a very simple environment in order to make some first steps in VDI. For this purpose, I've installed the View Connection Server and a Win7 VM on my ESXi 6 host.

Then I've verified the installation using the Horizon Client on my workstation as well as by using HTML on a different computer. That worked without any problems and PCoIP performance was as good as using RDP. But then I've bought an old Tera1 thin client (built into an LG display) and the results were ridiculous 😞

Yes I know that those Tera1 clients came into age but I can't imagine that people accepted this horrible performance in the past. I mean, sometimes the screen redraws only once a second and even keystrokes seems to be delayed that bad. And I'm only talking about basic Windows usage like navigating through folders, renaming files, ...

I've double checked that it isn't a network or ESXi performance issue. The performance counters of all affected devices / VMs are absolutely in range (no packet loss, PCoIPs network latency is shown as 0msec, no WAN/Routing involved, entire network contains only 8 workstations).

The display resolution is set to 1024x768 and I've set the VMs VRAM amount to 128MB.

So my question is if it's worth to try a different zero-/thin client or to drop this strategy completely and switch over to full-blown clients.

[mougT · Accepted solution]

Hi

No, Tera1 clients are not that slow. We still use a lot of Tera1 devices, and they have better performance than some older Windows computers that we have. It has to be either your network or the zero client itself that has a problem. Are you using the same network endpoint as your computer when you are testing the zero client? Have you updated the firmware on the zero client?

[blackcopy]

Thanks for making that clear to me. I really couldn't imagine that people have worked with these kind of lags for the past five years 😉

I'm using the latest available firmware for Tera1 CPUs (4.5.1) and yes, I can confirm that the network endpoint is okay. An old Asus eee-Book outperformed the Tera1 client on the same switch port without any problems.

Strangely I've just noticed that somehow now it working quite well at the moment. And for the first time, I see "real" values on the device's internal webserver for Roundtrip latency and Pipeline bandwidth.

The last days there was just a "0".

Maybe the device is really broken 😞

[glennvelsol]

I use Tera2 zero clients (Wyse P25) and performance is amazing. I tried a lower end thin client and performance sucked, so my advice is not to cheap out on thin/zero clients if you do roll VDI out.

So many things can effect your performance as well, there are a bunch of group policy settings that you can change for PCOIP that will greatly improve performance as well.

I originally piloted VDI on a old Dell server with built in storage...and performance was really bad.... I than redeployed to our production environment where the real deal hardware was and what a difference. I now use a extremeIO brick all flash storage and it's money! Your switches can also effect performance.

[blackcopy]

Okay, I'll take a look at those Dell clients. Our "roll out" would consist of no more than five clients, so I guess that an extremeIO card would be much too oversized 😉

Anyway I haven't located any kind of bottlenecks on our ESXi server at all. Neither regarding storage (which is a local SAS RAID1+0 in our case) nor CPU or memory resources (two v3 Xeons / 64GB).

In the end I guess my Tera1 device is broken and maybe it simply gets too hot after a few hours of work...