IT managers often say “let us not fix tickets as they come in, let us resolve the issue before the users are aware of it.”
Without the right tools, what ends up happening is that IT staff is always in “firefighting” mode, overwhelmed by user complaints and trying to decode the log files from a complex environment which appears as a “black box”.
When large VDI deployments are being designed and architected, lots of numbers get tossed around from compute requirements, storage capacity and IOPS, to user count, application set and type of user.
System integrators try to estimate requirements for the environment, but they are usually just that: estimates!
Once the project gets underway, the first item on the list is usually testing by a small group of pilot users. The environment is tweaked based on their feedback, more users are added, more changes are made, and so on. The last step is usually the scariest – open the floodgates and migrate a large population to the virtual desktop environment.
But what if the real-world demands exceed the estimates? What if an undetected configuration change blocked access to the deployment? Without a proper monitoring solution in place from day zero, it is hard to answer these concerns with confidence.
One of our Citrix Ready partners, Goliath Technologies, offer monitoring tools that provide deep insight into otherwise complex and overwhelming environments. We’ve noticed that by setting up the Goliath Performance Monitor (GPM) at the beginning of a project you will have historical data to allow staff to go back through history and see exactly when the disk latency went above 30ms average, or when the hypervisor host CPU became so overcommitted that the CPU wait times rose above 500ms, and so on. It is valuable not just to assess the sizing and design of the infrastructure, but identify “growing pains” to eliminate delays well before the virtual desktops go into production for majority of users.
Successful virtualization projects need end-to-end monitoring
A monitoring solution that only uses SNMP, WMI, and NetFlow may crack when watching over tens of 1000s of desktops on top of a hypervisor, on shared storage and intricate networking in the datacenter. A proper tool must integrate and gather data directly from all layers in the environment – including the hypervisor, the compute, network, and the control software orchestrating thousands of connections every day.
Monitoring tools that are built to monitor networks and Windows servers may not be capable of knowing which counters to use or how to integrate with Citrix specific APIs to align Citrix session behavior and performance with infrastructure conditions. Furthermore, technologies dependent on wire-data have no facility to provide the needed information because they have no mechanism to leverage the APIs or context from where the sessions take place – the servers.
Having the ability to understand, monitor, report and alert at every layer of a complex deployment empowers technology staff to know exactly what is going on in the VDI environment, when there is an issue, and where to start troubleshooting. For example, one can troubleshoot slow session performance by digging into the ICA Channel.
XenApp and XenDesktop is the best desktop virtualization solution for Citrix XenServer, Microsoft Hyper-V, and VMware vSphere. Goliath Performance Monitor is certified Citrix Ready as well as VMware Ready, which allows direct integration with vendor APIs to gather immediate and accurate data, and provide comprehensive reporting on the hypervisor as well as desktop/ app virtualization components all in one console.
Validate design estimates for compute and storage
Estimating the requirements for any virtualization project is difficult – especially when you are talking about thousands of end user desktops in the datacenter. These problems don’t show up during the pilot phase when there is only 2-5% test load. But as all users start to go live, a boot or logon storm during 50% system load may consume all available storage I/O.
The estimated IOPS may now require revision to account for peaks and bursts. Unfortunately, by this stage it is too late to find the budget to upgrade the hardware. Goliath Performance Monitor would notice if you end up with unusually high average datastore latencies as displayed below, and proactively alert if storage was not designed properly in the pilot stage.
Beware: changes break things
Multiple tweaks and changes are made during a proof-of-concept or initial deployment to fix issues or respond to user feedback. Unfortunately, all changes may not be run through a change control in the pressure of project deadlines, making it difficult to track and determine if a change was the true cause of an outage.
No one wants to see the dreaded ‘Cannot complete your request’ error because an incompatible change was made to Citrix Storefront or NetScaler Gateway configuration. The Goliath Logon Simulator will detect this issue immediately. By having a monitoring tool running from the beginning of the project, you will be alerted if the most recent change broke remote access or trace the root cause of crippled performance.
In the screenshot below, GLS is alerting us that application was not found:
Start early, monitor proactively
A monitoring product that can actively monitor the performance of the complete environment, not just the key components, as well as actively test logons, should be deployed from day one. XenApp and XenDesktop health check should be part of your proof of concept (PoC) and pilot strategy, not an afterthought. Not only does that proactively identify and fix potential issues in the pilot, but also accelerates the deployment and help guarantee successful project completion.
Try the 30-day free trial of Goliath Performance Monitor and Logon Simulator and see if it eliminates the “firefight” from your next project.