Load Testing in 2014: Project VRC Survey
Project Virtual Reality Check, the independent research body, today published its annual “State of the VDI and SBC Union” white paper. This publication contains survey results from over 800 participants and answers interesting questions about how enterprises use hypervisors, operating systems, antivirus and other desktop virtualization components. Download the white paper. In this blog, I will look closely at the question and very interesting responses on how companies load test.
As you can see, 43.6% of the survey respondents indicated that they use load testing to size and validate capacity, benchmark, stress test and, of course, to validate changes before implementing them in production. These are all common use cases for Login VSI and I have to presume these companies are doing their best to responsibly plan and manage their VDI/SBC environments.
What surprises me is that over 44% answered that they are currently not using any load testing solution. What’s more, the most common (23.2%) reason for not load testing reflects the outdated perception that load testing is too time consuming, expensive and resource intensive. While setting up a typical test bed used to take several months and require expensive software, the reality is that today these same constraints do not exist.
At Login VSI, our mission is to help make enterprises more successful in their VDI/SBC implementations. One specific way we are doing this is by changing how companies approach load testing. Over the past few years, we have been working on simplifying testing and improving test insights. As a result, an administrator can now set up a test within a few hours. This means you can now very quickly gain a tremendous amount of insight into your own environment and how it handles loads and changes.
Reasons why there are no more excuses for failing to load test:
- Turnkey workloads: No need for scripting
- Ease of use: Everything is wizard driven
- Scalability: Login VSI works in both small and extremely large environments
- Broad platform support : VMware Horizon View, XenApp, XenDesktop, Microsoft RDS, etc.
- Standard infrastructure: No need for special test hardware
- Easy-to-interpret VSImax test results: No need to analyze the data yourself
Another surprise in the survey results was that only 3% of respondents said that they are using load testing to validate changes in their production environment. That means that about 97% of companies are implementing changes to production environments without testing the performance impact. This is a serious risk. Any enterprise desktop virtualization implementation is build on a rather complicated platform with many dependencies that require different IT disciplines. Also, any successful VDI environment has the most changes to absorb as a single platform in the typical enterprise data center in comparison to others.
As an example: Last year’s Project VRC research showed that upgrading from Microsoft Office 2010 to 2013 costs about 20% capacity. Not planning for change will surely impact performance for end users.
Fifteen percent of this year’s Project VRC survey participants indicated that they will migrate to the next version of Microsoft Office within a year. Without testing the impact of this migration, these companies will run the risk of losing 20% user capacity. What is the cost to your business if 20% of end users suffer several days of downtime because of limited capacity? And what about all the other application updates? Or storage updates, broker upgrades or a simple monthly hotfix from Microsoft. Too many times trouble comes out of change in production and at scale.
Instead, Login VSI wants you to reap the benefits of high availability which is why we encourage testing and validating every minor and major change so you will never run into any unexpected performance issues. Simply put: find trouble before it finds you.