Manoj,
I've followed this thread with some interest, and have learnt something
from what Mike's been saying about how Guacamole handles image
compression etc.
I'm not able to contribute much from a softwarec perspective but there
are a couple of things that I wonder about.
In your tests it appears to me that, generally, the Guacamole <-> xrdp
traffic is much higher than Tomcat <-> browser, as one would anticipate.
Assuming your Guacamole <-> xrdp connections to be on an internal 1GbE
network then one would expect Tomcat <-> [external] browser experience
to be much quicker than say xrdp <-> [external] MSRD client.
In an earlier post you said:
My Observation is---
we observe that in my colleague company those people used Microsoft
remote desktop Web client (using activex) for 1200 connection in 10
Mbps for huge transaction. and
we used Xrdp+wine+Guacamole with 600 connection with 50 Mbps bandwidth .
what protocol they are used ( Microsoft remote desktop Web client )
is taking less bandwidth compare with Guacamole.
From this I was interested to see what information there was regarding
the bandwidth requirements for MS RD Web Client vs MS Terminal Services
Client.
I found this website article:
https://www.rdsgurus.com/microsoft-rd-web-client-html5-performance-testing-part-1/
Although not completely clear my take on their results is that MSRDWC
could use similar, or possibly more, bandwith than MSTSC (or it could
use ~1/2 in some cases). They explicitly state further research is
needed so the results should be considered provisional at this time.
*If* these results are in the typical ballpark then it would seem to me
somewhat at odds with what you said earlier, and with the results you've
charted.
To clarifiy.
(1) From the article let's say MSRDWC bandwith typically = MSTSC bandwith.
(2)You measure Guacamole <-> xrdp bandwidth significantly higher than
Tomcat <-> browser (let's say this equates to what we'd expect typical
MSTSC bandwidth to be).
(3) Extapolating; your colleague company is using MSRDWC, therefore with
no other changes or tuning we might ordinarily expect their bandwith
requirements to be higher than yours since, from your's and Mike's data,
the Tomcat <-> browser bandwidth should effectively be less than MSRDWC.
(4) Yet you've said they have twice the connection numbers with 1/5
available bandwith, and although not stated the intimation is that their
user's experience could be better than yours?
Now I realise I'm drawing a fairly long bow, and making some pretty wild
assumptions based on possibly erroneous data, but at this point the
comparision just doesn't add up.
There are many possibly variables that might explain this but I wonder,
initially, if there are some other differences in what service your
colleague company is delivering compared with yours? For example is
theirs a much reduced colour depth, is it limited to a specific app with
little screen updates, do they have burst mode data capacity, do they
have fewer _concurrent_ users etc?
Ultimately Mike has said several times that you simply need to allocate
more resource for what you're doing, but it may assist yourself and the
community if you could obtain a similar bandwith log from your colleague
company as you have for yours. It would be good if the data were
standardised as much as possible (ie. perform exactly the same desktop
tasks) and the same colour depth settings etc were utilised.
If this were effected I think we'd have a much better idea as to the
bandwith requirements of Guacamole vs MSRDWC. From this one might then
know if there's any real (comparative) issue between your service and
that of your colleague company, or not. It could also give some
potentially useful info around Guacamole/MSRDWC performance...
On 3/03/2020 4:06 p.m., Manoj Patil wrote:
What tool u used for measuring bandwidth.
Is there any resolution?
As per your snanshot RDP take to much bandwidth utilization . if u
also 600 active connwction then the bandwidth utilization is around
40- 45 mbps.
On Mon, 2 Mar 2020, 00:45 Mike Jumper, <[email protected]
<mailto:[email protected]>> wrote:
On Sun, Mar 1, 2020 at 1:59 AM Manoj Patil <[email protected]
<mailto:[email protected]>> wrote:
Dear,
I am deployed the Microsoft environment at my end and measure
the bandwidth data send and received.
using wireshark.
please find the attached file one is microsoft web rdp
bandwidth calculation using wireshark and second snapshot is
guacamol server calculation using wireshark .
If you want to compare bandwidth usage reliably, you will need to
measure and compare the two sides of the same session: one
measurement being the browser <--> Guacamole traffic and the other
being the guacd <--> RDP traffic. For example, here's my
statistics for the first week of December last year:
glen-demo-stats-2019-12-01-through-2019-12-07.png
The graph shows total Guacamole bandwidth usage (green line)
against RDP usage (orange line) for the same servers across all
sessions. The purple line is the total number of active sessions.
In general, the two bandwidth lines follow each other, however
I've always observed the RDP line to be significantly higher,
presumably due to using poorer image compression. The only times
I've seen the Guacamole line peek (slightly) above the RDP line
are when there is extremely low activity.
If you are absolutely sure that you are measuring effectively the
same sessions, connecting to the same RDP server, and that you are
using the same display size, performing the same actions, seeing
the same graphics, etc. between them, I'm not sure what would
account for your measurements showing the opposite behavior.
- Mike
