I just finished up a monitoring deployment of our IBM’s Citrix XenDesktop monitoring solution in a VMware/Citrix VDI environment. The customer is monitoring their VMware and Citrix XenDesktop enviroment with IBM’s IBM Tivoli Monitoring based solution. I wanted to share a few key thoughts from the customer and my observations.
First, it’s key to have monitoring in place to monitor the entire Citrix XenDesktop environment including the license server, broker controllers, etc. But, it’s equally important to monitor the VMware environments as well as the key infrastructure components such as the storage and networking components. There have been incidents where problems with the back end storage affected the performance of the virtual desktops. In addition, there have been cases where the applications being accessed via XenDesktop such as their E-mail system was suffering performance problems. The end user assumed that the problem was related to the VDI solution even though the problem was the E-mail system. Therefore, it’s critical that the VDI team have monitoring visibility into all of the key applications that the users are accessing.
From a VMware perspective, the customer had good monitoring and reporting in place to detect problems with the VMware environment. But, they were just starting to look at capacity planning for he VMware environment. We discussed the benefits of IBM’s capacity planning and optimization tools. Customers can do typical what-if capacity planning scenarios, but can also optimize the environment to get more use out of the existing hardware and potentially eliminate servers along with the associated power and cooling costs, administrative costs, VMware licensing costs, etc.
In terms of XenDesktop, there were some key elements that were very important to the customer. The monitoring tools couldn’t have a negative impact on the XenDesktop environment. With some architectural changes made to the monitoring agent, we saw very low impact. Overall, the Broker Controllers were utilizing around 5% CPU utilization even with the monitoring. The environment is expected to grow, but with the new architectural changes, adding more concurrent desktop sessions won’t affect how much load the monitoring tools place on the Broker Controllers. Reporting was key. The product provided some key out of the box reports and the customer was easily able to use the Cognos based reporting tool to create a custom report for the VDI team. The report showed the historical utilization of various users of the VDI system over time. End users are being asked to use VDI rather than traditional laptops and the customer has the need to determine who is actually using the VDI environment. The out of the box thresholds provided most of their needs, but the customer created a few custom thresholds. One of the most important thresholds was to determine the percentage of available desktops within a desktop group are in use. Another desired threshold was to identify sessions that have been in an Unregistered state for a significant period of time. After a number of hours, the sessions associated with disconnected servers should become inactive, but that doesn’t always happen. That leaves the sessions in an undesired state. Through the user interface, it’s very easy to see sessions that have been unregistered for an extended period.
I will continue working with the customer and their VDI team. They are going to identify their top 10 most common problems and we’ll make sure that monitoring and reporting is in place to identify and resolve the problems. It was a great and educational visit.