Proactive management of end user experience using Goliath Performance Monitor and Goliath Logon Simulator
and XenDesktop session was established. The last thing I showed you was the
result of the Logon Simulator from the console. If you missed that article, it
is a good start before reading this article, you can read the story here.
In this article I will go deeper in the product and show you how you can see
very cool and wanted metrics within the Citrix ICA and HDX channel, and how you
can monitor and debug a Citrix environment.
If we were to take a poll on the coolest report Citrix EdgeSight used to have,
I’m sure #1 and #2 would be Logon Details and Session Startup Details, which
gave very detailed information, step-by-step client side and server side. Logon
Details gave insight in GPO load time, logon scripts and so on.
combo, but they were hidden in a license that was hard to sell and now Citrix
decided to kill all cool features to start fresh, leaving you and me thinking
of the past.
No worry, those two cool reports are still there, not delivered by Citrix but
through Goliath Technologies. When you do proactive monitoring with the Goliath Performance Monitor the agent
measures all this data and will report this to the console. The great people at
Goliath added this report in the console so life without EdgeSight is wonderful
Enough talk, let me show you what they have in the product.
while monitoring a Citrix environment are:
- ICA/HDX Latency
- GPO load time
- Bandwidth and congestion
- Connection time to the broker/web front
- User profile load
- vGPU performance (becoming more and more common)
Goliath offers several reports that will help you
understand the end user experience. Just to remind you, Goliath performs simulated
Citrix XenApp and XenDesktop user logon with actual user accounts. With that in
mind you can understand that loads of data is gathered.
debugged in details afterwards if needed through a search function. I started a
couple of sessions over a few days to get some data.
In my last article
we primarily looked at the reporting of the Citrix Logon Simulator
section, now we focus on the section XenDesktop/XenApp session which is much
more interesting for IT guys. In the report below you see two sessions that were
active, if you look closely you see two important metrics right in the open.
The logon time and the ICA latency are reported already, giving you
instant information about the user experience. So were done, right… or wait
there is more, so much more.
see more data you click on the + mark next to the session, and you get
into the reporting part. A number of tabs and sub tabs will open where you
find detailed information about the session, the server and the user. First we
start with the logon duration that was shown also in the main screen.
The logon duration is more than just logon, it is made up of several steps
that together make up the logon time. The logon time is made up of the
- Logon time
- Client validation & brokering
- Scripts that run
- User profile
All are shown in the report and the time they take, also per step a detail is given.
If you look at the details in the report above you see that it shows
exactly what server, domain, etc. is talking to. This is the kind of detail you
want in a report when you debug slowness,
If we look at the break down of all of this, what I used very much in EdgeSight
was the Start Detail Report because it
showed all aspects of the connection. If you never worked with EdgeSight you
might think it a silly name with lots of letters but they all actually stand
CASD for instance is Credentials
Authentication Server Duration. In plain English, that translates to how
long it takes to get your credentials verified. On the client side you have
LPWD. It’s not a police force, it’s Launch
Page Web Server Duration, which means how long does it take to launch the
web page (Storefront or Web Interface).
Depending on your environment you get different values and more or less values.
There are lists on the Internet to tell you what the abbreviations mean,
I’ve included one that was written for Edgesight.
A detailed report about the ICA Channel delivers
metrics that are important to understand the end user experience. The ICA
channel performance will be different from normal network performance. Latency for
the ICA channel and network latency can be different.
very important metric to be measured. Together with the bandwidth and network
latency, this gives a good indication whether the user is having some issues.
Latency is the time it takes for a package to travel from one point to it’s
destination. The higher the number, the worse your user experience gets.
Next to latency there is available bandwidth, not to be mistaken with speed.
The available bandwidth has a direct influence on the latency, so if your pipe
gets congested, latency goes up.
This has nothing to do with the speed of your connection. You can have a
100M connection but when it’s congested, it will not perform. To solve
congestion you add more pipes. Unfortunately that doesn’t mean latency will go
down at the same rate, latency can have more influencers.
Goliath offers a report that shows these metrics together so you see the relation
virtual machine and see numerous metrics that will help you understand the
performance of the user session and the machine it lives on. This virtual
machine metric dashboard has many other tabs showing relevant
information about the server. The following tabs are shown:
- CPU and Memory
- Disk use
- Disk Peformance
- Disk Space use
- Network Throughput
- Storage & VM Limit
- Agent Up&Down
- Graphics Metrics
- Storage Trends
- Storage Allocation
- Storage Usage
Well if that doesn’t give you enough information I don’t know what will, let’s take a look at some of them.
If we look at server disk performance, we see
the performance of the disks, throughput and IOPS being displayed. In my lab
with one or two users there is no issue but in a live environment this might
be valuable and show if the user is having issues because of a lack of
A minor report is the Disk space use which allows you to identify conditions where there
is unchecked disk space growth on your diff and identity disks.
Network throughput is important in solving issues
with end user experience. If for some reason your network speed is low, it
might impact the end user experience.
As I said before speed is not the
most important factor in end user experience. but a lack of speed also doesn’t
help. All metrics work together to deliver the best performance. Think about
the E1000 versus the vmxnet3 NIC and the performance difference it gave.
One of the cool features I think is the ability to
monitor the usage of vGPU with their Performance Monitor for NVIDIA GRID vGPU. Goliath is working closely with NVIDIA to capture the data of the
cards that need to be captured. I borrowed some screen shots from Goliath because
my home lab is not equipped with a grid card. If you monitor an environment
without a vGPU this tab won’t be visible.
Metrics you find here are;
- Frames per second
- vGPU Compute Utilization
- vGPU Memory Utilization
When debugging or monitoring sessions you need
to know the context of the issue the user is experiencing. The context of
the user experience is to know which other processes are running and using
resources. The application performance report is doing just that; it shows the
other processes running with metrics per process.
A good start is always the overview of the session
you’re debugging. Without a good sense of the session the metrics say nothing.
Together with Citrix Director, which does a good job of showing information
about a session, and this report also shows the session properties.
Here you can verify versions, networking details and so much more, so I think
it’s a good starting point and a good report to end this article with.
I think Goliath offers a very good tool to do
simulated Citrix logons with actual user accounts and that the reports
are useful to monitor and debug a session. If you look for
a monitoring tool to look at the user logon performance of your Citrix
environment, Goliath would certainly be able to fill that gap.
The addition of vGPU is something that will be needed more and more, even
Citrix XenApp servers will get graphic cards in the near future.
versions of the tool and as I wrote in my previous article. Starting an
application is something that might be useful because that is a user action
that might also vary in time. If they could start an application (e.g. Word)
and report time it takes to load certain plugins, that would be useful.
Another addition would be to be
able to do exactly the same as they do with Citrix in a VMware Horizon View
environment. It all depends, of course, on VMware opening up the API’s to allow
them to ge the data from the channel. In my job, I see a 50/50 in deploying
Citrix and Vmware, and for VMware there are less options to do simulated
Perhaps integration with vCops, or
as it is now called, vROps (vRealize Operations) would be interesting because all
data is collected there already although not all our customers have that
If I had to make a choice, it would be both ways. Integrate and be independent
as well to accommodate both customers.
Hopefully, my article gives you a better idea of what Goliath is capable of. To
see for yourself, try a 30-day
free trial or demo of Goliath