Recently I was working with a client who had experienced an unexpected power outage. Once the power returned, the PCs had to be rebooted. The client uses Citrix to access Millennium, and the rebooting overwhelmed the Citrix Zone Data Collectors. The overload delayed system access for everyone trying to get back into Millennium® — even though the data center stayed up during the outage.
Was the overload inevitable? Or would it have been possible to tune the site’s five-year-old hardware to better recover from or live through abnormal spikes in activity?
These questions are important because many sites use Citrix’s XenApp servers to deploy Millennium. Citrix — how most users refer to all of the various Citrix components — allows a small set of servers to present Millennium applications, which individual PCs then access by connecting to the Citrix servers. The Zone Data Collector (ZDC) is used by Citrix to evenly balance the PC connections across all of the XenApp servers. Without the ZDC, the first Citrix server in the group or zone would get all of the connections, which would quickly overwhelm the server and cause other issues.
As my client discovered, even with the ZDC, abnormal events can overwhelm the server. So I want to suggest some basic tuning steps to help you balance the ZDC loads and ward off future overloads. We’ll discuss the data store (often SQL Server) tuning and registry changes that can make the data store and ZDCs able to handle sudden increases in volume.
I would suggest you review your Citrix configuration even if it was set up by an “expert.” Often the experts are skilled in following a specific set of requirements to set up an environment or domain and show you how to do it next time. Then they leave. The more knowledge you have about your environment and its configuration, the better you are positioned to address potential issues. Ask yourself, “If I had to accommodate a doubling of connections or clients in less than 60 minutes, where would I have problems?” After reading through the steps I recommend, you can create a list of what tuning needs to be done and start the change control process to put these changes into production. That way, the next time you have a power outage at your organization, the recovery time will be shorter and less painful for you and your Millennium users.
For details on the Citrix tuning process, click here.
Prognosis: By moving your data store to an optimized disk partition and tweaking some registry settings, you will find that Citrix can become much more responsive, even when the workload more than doubles.
