In the virtualized server environment, fixing performance issues can quickly become a full-time job. The virtual environment is different than the physical environment because of its many layers that abstract the potential causes of a performance problem.
In our webinar Managing Performance In The Virtualized Environment, I discuss the two options that administrators have to resolve performance issues in the virtual environment. They can either try to fine tune their way through the problem, or they can throw hardware at the problem. Of course, that is the same set of options for almost any performance-tuning task in the data center. What makes the virtual environment unique is the theoretical simplicity and high payoff of throwing hardware at the problem.
The virtual environment has multiple layers of abstraction, and getting a holistic view of these layers of abstraction can be challenging with the tools that come with the infrastructure's various components. These tools are not aware of the other components and how they interact, so they each report on themselves, not on the whole. Manually consolidating the information from each of the tools to create a holistic picture can be time consuming, error prone, and ineffective.
The difficulties of managing VM performance lead you to investigate hardware solutions as a bandage to performance problems. Solutions like solid-state drives, faster networks, and server-side caching are all attractive options and in most cases do provide an immediate, but potentially temporary, performance boost.
Every virtual environment can potentially benefit from faster storage and networks, but hardware that improves performance comes at a premium price. Blindly throwing this hardware at your virtual performance problems results in buying excess performance, which goes unused and wastes your IT budget.
As we discuss in our article The VMware Memory Balancing Act, there are changes that you can make to your environment that can improve performance and require no additional hardware or software, just visibility. Great examples are RAM utilization, as discussed in that article, and proper alignment of partitions. A better set of analysis tools should provide both.
Holistic View Is The Key
The virtual environment has too many moving parts to try to manually consolidate multiple performance-monitoring tools on a spreadsheet. Instead, you need tools that can provide a view of the entire environment, showing how the different components of the virtual infrastructure interact with each other.
Not only will these tools help you get more mileage out of your current infrastructure, they will make sure you buy the right hardware at just the right time for the maximum return on investment. The truth is, in a maturing virtual infrastructure, you will eventually need both faster hardware and the ability to tune and maximize that hardware.
Follow Storage Switzerland on Twitter