Measuring Load Speed on Thin Client (VNC Stuff)

Toggle sidebar Toggle sidebar
B

Balkoth

Reputable
Jul 28, 2014
2
0
4,510
This may be suited to the Applications forum more but since I'm trying to measure something regarding a thin client + server network...

I'm trying to measure the relative performance of different encoding schemes across different network speeds and packet loss. One of the main ideas of measurement I have is measuring the time it takes for the following to happen:

1. Press enter on the thin client which sends data to the server and the server starts an MRI viewing application
2. Server then does all the processing work and sends the screen image data back in a preselected encoding scheme
3. The client "unpacks" the data and fully loads the image

So basically, the time it takes from the time you hit "go" to the time the image is loaded on the *client* side.

I have some ideas I'm looking into with for the "go" timestamp, ranging from just trying to timestamp on the client or timestamping when the server receives the message (and subtract off the latency and compensate for different internal times potentially), but I'm feeling stumped about how to track when the MRI image "loads" (I say "loads" because it's not a static image but rather a rotating display that can be manipulated).

Thoughts on how to do that? Or alternative measures of performance? I've been using UltraVNC for the various encoding schemes so far but using TightVNC or other VNCs is also possible.

Thanks!
 
Sort by date Sort by votes
popatim

popatim

Titan
Moderator
Dec 2, 2009
38,898
978
135,890
Off topic question for you
What monitors are you using?
Whats the hardware on the client end?
I ask because we use some expensive full gamut, calibrated monitors with each individual workstation doing the processing with quadro cards and xeon processors. Which gets expensive. Our radiologists wouldn't put up with much delay either so I'm interested in how well this works for them and if they are satisfied with the image quality.
Thanks :)
 
Upvote 0 Downvote
B

Balkoth

Reputable
Jul 28, 2014
2
0
4,510
In theory, a variety of monitors and hardware.

In practice, have several desktops and laptops with various operating systems that could be used for the testing, some better than others. None of the hardware we're currently using is fancy (see below for reasons why) but if necessary we should be able to get access to better hardware for our research purposes.

We're working with Ohio State University's RIVVIR system (https://labs.bmi.osumc.edu/lailab/projects/rivvir/) currently and doing some preliminary research on the last sentence there: "We are currently focused on executing performance analysis studies on RIVVIR and determining areas of improvement for users of the system." In my lab's case we're currently looking at various encoding schemes to compensate for various network situations (specifically poor networks with low speed, packet loss, etc) along with multi-user performance testing right now. Still somewhat preliminary.

Subjectively we can simply induce poor network connections via emulators or going to locations with known poor connections and simply "gauge" how it performs with different schemes...but we're also trying to find a good objective measurement for the different encoding scheme performances with different networks. We may need to resort to simply checking bandwidth used by each scheme to determine what schemes RIVVIR users should use but we were hoping to come up with a better objective measurement if possible.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom