A recent report tells us that over the last decade, Citrix technologies have advanced significantly. From being a provider of remote access solutions, the Citrix portfolio of solutions has dramatically expanded to include Citrix XenApp, XenDesktop, XenServer, XenMobile, Sharefile, and so on.
That same report finds that Flexwork and telework, bring your own device, mobile workspaces, PC refresh alternatives, remote partner access, etc. are just some of the ways that organizations are using Citrix technologies. The key role that Citrix technologies play makes performance management for these technologies and their supporting infrastructure critically important. In most cases, these technologies can be successfully deployed only if the user experience is satisfactory.
To deliver the best possible user experience, Citrix environments need to be not only well architected but also well monitored and managed, to identify and diagnose problems early on, and prevent issues from escalating and impacting end users and business processes.
Citrix XenApp and XenDesktop infrastructures are also difficult to troubleshoot. These infrastructures involve multiple tiers of hardware and software and each of these tiers run on virtual platforms. Furthermore, each of the tiers involves technologies from different vendors and if there is no single management solution available that provides end-to-end visibility, most often problem diagnosis is a hit or miss process.
Users are not aware of the complexities of the underlying infrastructure, nor do they care. What they need is for their application and desktop access to be fast and reliable. When they see a slowdown, the complaint is always that "Citrix is slow". Faced with user complaints, administrators have to look at different tools (possibly one for each of the tiers) and do a manual analysis to figure out where the problem lies: Is it in the network, storage, virtualization platform, Citrix, or the application tiers? This manual process can take a lot of time, leading to unhappy users.
In this paper, you will learn how you can address the top five most common Citrix XenApp and XenDesktop performance challenges.
This is easier said than done. Citrix XenApp and XenDesktop are two of the most performancesensitive elements in an enterprise network today. Web applications, which get a lot of attention from performance experts, are often more tolerant to slowdowns because they are not session-based. Citrix XenApp and XenDesktop accesses are highly interactive and session-based. A small glitch in the network any time during a user's session can result in the session getting disconnected, forcing the user to lose the applications and data they were working on and to re-login to the desktops and applications.
Users call the helpdesk about a problem. Their complaints often relate to the Citrix XenApp or XenDesktop tiers because that is what they see when they login. The helpdesk escalates the issue to a Citrix expert, who manually analyzes data collected by many tools to determine what caused the problem.
There are several problems with this scenario:
The more user complaints the helpdesk can resolve without involving Citrix experts, the faster the problem resolution and the lower operations costs (experts can perform more productive tasks than firefighting).
In a recent survey of 500 Citrix professionals conducted by DABCC and eG Innovations, 78% of respondents indicated that less than 50% of the time, the helpdesk was able to resolve user complaints without involving Citrix experts. The results indicate that there is a significant scope for organizations to optimize the way that they are doing Citrix performance management. By leveraging tools that empower the helpdesk to do more, organizations can reduce the workload on their Citrix experts, lower their support costs and improve user satisfaction.
The solution to this problem is to empower the helpdesk with a comprehensive, unified view from where they can see the key performance metrics for each user's session. All the helpdesk personnel have to do is type in the user's name and they can drilldown to a view of the current performance of that user's session. Figure 1 shows the performance dashboard for the user. From the dashboard, the helpdesk can also see the current alerts for this particular user.
One of the key indicators of user experience in the dashboard is the user experience rating is low. The user experience may depend on many factors – the login time for the user, application launch time, ICA roundtrip latency, CPU usage of the user's processes, etc. The user experience rating is a single metric that the helpdesk can look at to determine if there is a problem with the user's session. This metric is derived from of the other metrics. If the rating is 100, the user experience is good. If this rating is anywhere below 90, the user is experiencing performance problems.
From the dashboard, the helpdesk can also drill into the server that the user is connected to and analyze other metrics about the server health to determine if there is a server-wide problem that could be impacting the user experience.
Figure 1: The Citrix User Experience Dashboard for a user kevin
From an end user perspective, the advantage of this approach is that you either get your problem solved on your first call or you learn that someone is working on solving your problem and this leads to greater user satisfaction. As a Citrix administrator, imagine how much time this is going to save vs. having to troubleshoot every single user call.
Several years ago, Citrix experts, Brian Madden and Kevin Goodman created a chart describing in detail the steps involved during the Citrix logon process. This chart had over a hundred different actions and steps.
Given the large number of steps involved in the logon process, it is no wonder that slow logons are the most common complaints received by Citrix administrators (48%). From an administrator's standpoint too, this is one of the biggest, most complicated problems for them to resolve. The problem is aggravated when users connect to the Citrix infrastructure from multiple devices.
Citrix administrators need access to a monitoring tool that tracks every single user logon to the server farm and can help quickly identify why the logon was slow – i.e., which step was slow – was it authentication via Active Directory, or was it due to the profile loading time, or was it because of group policy processing, etc.
Figure 2 shows the breakdown of a user's logon time. A simple breakdown like this with thresholds applied to each of the metrics collected makes the Citrix administrator's job in troubleshooting logon slowness very easy.
Figure 2 – Why is Citrix user logon slow? A breakdown of the logon time helps answer this question
A typical Citrix infrastructure has software and hardware tiers that have to interact and work together to deliver the service effectively for the end user. So when a user calls to say 80% of Citrix that their application didn't launch administrators quickly enough or is sluggish, the are using issue could be anywhere within multiple tools those tiers. It could be a database to manage the server slowing down or a network performance switch having a problem, but for of their your user, it's just going to look to infrastructure. them like they are having a Citrix problem.
Right now, you have information in silos – you have to look at VMware vRealize Operations for the virtualization platform, EMC tools for storage, Cisco tools for the network, Windows Task Manager for Windows operating systems and Citrix Director/EdgeSight for Citrix XenApp/ XenDesktop.
What every Citrix administrator wishes for is a single pane of glass view of their entire Citrix infrastructure – from end-to-end. From this unified view, they can see how each and every tier of the infrastructure is working. When there are problems, they are alerted to these problems in advance. Plus, if a problem in one tier affects other tiers, they can clearly see from the monitoring console where the cause of the problem lies and where the effects are. This way, they can focus their efforts on the cause of the problem and not be distracted by the effects.
Figure 3 shows a topology of the Citrix service infrastructure. Color codes highlight the areas that are working well and ones that are not. Drilldowns provide more details of the problems.
Figure 3 – Topology of a Citrix XenApp service highlighting the different tiers involved and where the more severe problems lie
Instead of chasing multiple different issues, with clear color codes in the service topology, you can identify the real problem and to resolve this quickly.
This is a relatively new issue that has cropped up in recent years. More and more applications are becoming Webbased. Citrix administrators are publishing browsers and making them accessible to users. When they see a problem, all that the user can tell is that the browser is slow – the question is whether this is because of a Citrix server bottleneck, or because of the browser issue or whether it is because the backend application being accessed is slow.
To address this problem, Citrix administrators need access to monitoring solutions that provide visibility into browser activities. Besides viewing standard server health metrics, administrators need to know which URLs/websites did a user access when the problem was reported. Furthermore, they may also want to know what other users of the Citrix server were doing at the same time and what URLs/web pages were they accessing.
Figure 4 shows the drilldown that tells what applications are running in a user's session. Notice from the CPU Time column that one of the Internet Explorer processes is taking 85% of CPU time. One process taking up so much CPU on a shared Citrix server is not normal or acceptable. The Website Title and URL columns illustrate what the user is viewing with Internet Explorer. In this case, the user is accessing YouTube (a non-corporate website) while at work, taking up a lot of the CPU available on the XenApp server and possibly affecting the performance that other users are seeing. Such deep insights into browser access on Citrix XenApp server and XenDesktop VMs is a requirement these days.
Over 68% of Citrix professionals surveyed indicated that being proactive was their biggest need. Citrix administrators want to find problems before end users do, and as much as possible, they want to avoid problems altogether.
A Citrix infrastructure has many tiers, each of which uses different protocol layers. Hundreds of thousands of metrics can be collected at each layer and every tier of the infrastructure. This is too much data for an administrator to manually analyze. What Citrix administrators need is a monitoring solution that can analyze the data and present actionable information to them.
One of the ways of doing this is through auto-baselining. The monitoring system understands the time of day, day of week behavior of the metrics collected and intelligently sets thresholds based on which alerts are generated. For example, in your environment, a backup job may always run at 2AM. As the monitoring solution learns that you run backups at 2AM, it will not raise an alarm at this time even though the server's disk activity may be high.
Figure 4 – Drilldown into a user's session activity showing the web URLs that the user is accessing Figure 5 – Auto-baselining of performance metrics is critical for monitoring to be proactive
common Citrix XenApp and XenDesktop performance challenges and provides a total solution for performance management of Citrix infrastructures. With this unified solution, Citrix administrators can track the end user experience, understand the service performance and drilldown to see the performance of each of the individual tiers of the infrastructure. Performance correlation and analytics embedded in eG Enterprise help Citrix administrators to be alerted preemptively to problems and to diagnose rapidly where the root-cause of the problems lies. Analysis of historical performance built into the solution provides key insights for right-sizing Citrix infrastructures and for effective capacity planning.
Figure 5 – Auto-baselining of performance metrics is critical for monitoring to be proactive
Figure 5 above highlights how auto-baselining works. The baseline mirrors the changes in the data values over time, so thresholds are time-of-day based. This capability allows the monitoring system to be automated and proactive so administrators see early warning alerts that help them take control of the situation before the service is impacted.
The eG Enterprise solution from eG Innovations addresses the best practices we have outlined above to solve the
Figure 6 – eG Enterprise provides a total performance management solution for Citrix environments
eG Innovations is dedicated to helping businesses across the globe transform IT service delivery into a competitive advantage and a center for productivity, growth and profit through award-winning performance monitoring and management solutions. Many of the world's largest businesses and mission-critical environments trust eG Enterprise Universal Insight technology to enhance IT service performance in physical, virtual and cloud environments, increase operational efficiency, ensure IT effectiveness and deliver on the ROI promise of transformational IT investments.