Now, this is very cool! If you have been following my blog you probably noticed a previous article where I explain how to use a tool called Splunk to collect, index, search & analyse PCoIP log files across the entire organisation. On that same article I mentioned that I soon would be publishing a App for Splunk. I have been working with the Splunk team (Thanks to Brett Matthews, Tat-wee Kan and Daniel Miller) to create this app and it is now ready for use and distribution.
This app that you are about to download will allow you to monitor passively or in real-time all PCoIP traffic from your VMware View environment independent of how many concurrent users – anything between 1 and thousands of desktops. Yes, I said Real-Time!
If you already have some type of network monitoring in place, great, that is good practice and you should do it. However that will not give you anything near the amount of information you are about to collect and analyse.
Then you will say, yeah but I already have a zillion of log files to go trough – I don’t want one more. And I say – there are no logs to go through if you don’t want to, and everything will be displayed in a nice and neat dashboard interface that was created specifically for PCoIP.
Amongst the values you will be able to graph and trace are: RTO, Plateau, Limit, Latency, Loss, Variance and even CODECS in use. Yet, it is possible to segment by time, date, virtual desktop, or watch it all in real time.
These numbers will let you know how exactly PCoIP is performing in your network and how the users are perceiving their experience. I’ll soon write an article on how you understand and analyse each one of those values, but now enough talking and let me demonstrate what I am providing here.
Below is the dashboard or main screen where you can follow the PCoIP perceived packet losses with the Minimum, Average, and Maximum values. The dashboard is also tracing and graphing the Packet Loss across all virtual desktops. If you want to select a specific desktop or pool of desktops just type Source=”desktop_name*” in the formula field on top of the dashboard.
Additionally, the dashboard is graphing the Plateau Distribution, the Average and the RTO distribution over time. (Didn’t fit on screen so you can’t read properly).
We also created a set of pre-configured queries to make easier to search for common event but you may configure you own queries. If you want to see all the events just select “VMware – PCoIP Event Listing”. On the right side where you see a drop-down menu with “All time” you can select pre-defined periods such as the last 15 minutes.
The picture below demonstrates the RTO distribution over time and the tool allows you to drill down into events, sessions and fields. It’s really flexible and you know what – It’s FREE for database indexes under 500Mb.
Once you download and install the Splunk server package (download here) you will need to download the App from my blog (click here to download). To install the app just unpack and copy the folder and files to your Splunk_home/etc/apps directory.
You are now ready to process PCoIP log files. If you are still unsure how to find the log files or how to read them read on my previous article How to troubleshoot PCoIP performance. The App comes with few log files pre-loaded (log into Splunk_home/etc/apps/vmware_pcoip/logs) and you will need to download few more logs from your virtual desktops if you want to see what’s happening in your environment.
If you prefer to automate the log file download from your virtual desktops (I guess this is the nirvana scenario) then you will need to configure Splunk server as a Receiver and install the Splunk ForwarderLight in your virtual desktops. I would recommend installing the Forwarder in your ParentVM and recompose the whole lot, if possible. It’s very light and will not add overhead to your VDI environment.
For instructions on how to implement Receiver and Forward refer to the Splunk Administration Manual.
This is an initial dashboard and I already have some ideas of what I should change and implement for a next release. But if you decide to invest some time and create your own dashboard or change this one, please share with us.
I told you this one would be Cool! My Christmas gift to all View administrators
Some more screenshots of possible graphs: