How important are benchmarks in performance monitoring? How often should we run benchmarks, and how often should...
By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
we check actual performance against the benchmarks?
Data centers are complex and often dynamic environments where computing resources and workload demands are always changing, so workloads that are functioning properly today may not run equally well tomorrow. The problem is, in order for IT professionals to understand when something is "wrong," they must be able to compare it to something that is "right." Any differences provide important clues that can minimize wasted time and effort in a blind troubleshooting process. IT professionals use benchmarking tools and procedures to establish performance and utilization benchmarks when physical and virtual systems are operating in an acceptable manner.
When trouble strikes (perhaps identified by a user complaint or an automated threshold warning), a technician can use the same performance monitoring tools to compare current performance or utilization characteristics against the previous benchmark data. The difference in performance data can often lead to a fast and confident resolution of the problem. For example, if a baseline indicates 70% system memory utilization, but the server is performing poorly, with memory utilization at 90%, a technician can quickly try workload balancing to free memory.
In addition, performance monitoring can help organizations plan for capacity growth by watching resource utilization trends over time and recognizing the potential for future resource shortages that may precipitate problems. This allows organizations to avoid workload performance problems or disruptions by taking pre-emptive action to stop the problem before it starts. But this also requires a baseline and periodic benchmarks to see trends over time.
Benchmark frequency is harder to define because it really depends on the importance of the workloads. As a minimum, benchmarks should be used to establish a performance baseline when a system is first configured, put into service, workload balanced and vetted. If the system, workloads and usage patterns remain unchanged, it's generally not necessary to take new baselines (but periodic checks can reveal usage changes).
However, if the system received hardware changes or upgrades (such as more memory or a new network interface card), if the workloads are changed or upgraded (such as a new application release), or if usage changes (such as significantly more users or customers) it is absolutely appropriate to check performance and establish a new baseline. The faster or more dramatic those changes are and the more important those affected workloads are to the business, the more often a benchmark should be repeated. If in doubt, run the benchmarks more frequently rather than less frequently.
Performance monitoring and benchmarking are perhaps even more important for a virtual server than for a physical system. Remember that virtualized servers are almost always consolidated, and as more workloads vie for limited computing resources, the potential for resource shortages and contention becomes far more serious (and harder to detect). Pick the right tools to provide an accurate view of virtualized systems, and benchmark every VM to determine its performance characteristics during normal operation. This will speed troubleshooting and simplify future capacity planning.
Dig Deeper on Virtual machine provisioning and configuration
Stephen J. Bigelow asks:
How often do you collect performance benchmarks for your virtual machines?
0 ResponsesJoin the Discussion
Related Q&A from Stephen J. Bigelow
Not all hypervisors and OSes can play nice with others. The best method for avoiding complications down the road is to thoroughly test your nested ...continue reading
Theoretically, there is no limit on how 'deep' a nested VM can go. The best way to avoid any confusion is to appropriately label hypervisor nesting ...continue reading
Interested in giving nested virtualization a try in your data center? There are a few software and hardware requirements you'll have to meet first.continue reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.