To monitor performance, users look at the application itself. Various tools exist for monitoring application performance. However, in this recipe, we will use VMware vRealize Operations Manager (VROPS).

Tuning an enterprise application for performance is challenging because, to begin with, the host running the VMs in which these applications run might be dealing with workloads of varying characteristics. As a result, the root cause of a performance problem might end up being a combination of problems that are CPU-related, memory-related, networking-related, and storage-related.

As a system administrator, you measure performance in terms of resource utilization. From the perspective of the user, the best measurement of performance is the response time of the application itself. In addition to resource utilization metrics, there are two performance measurements that can help an application owner determine whether or not an application's performance is acceptable:

• Throughput
• Latency

Throughput refers to the number of transactions that can execute in a given amount of time. There are various ways to measure throughput: megabytes per second, IOPS, transactions per second, or instructions per second.

Latency is the measurement of how much time it takes to do a specific job. Examples are instruction completion latency, network latency, disk I/O latency, and response time. Throughput and latency are often interrelated.

Application performance issues are not always related to resource utilization problems. They can be caused by the application itself, such as malformed queries or lengthy SQL coding. In addition, application performance issues remain the same whether you are using physical hardware or virtual hardware.

VROPS provides visibility from the application to the storage and provides administrators with an overall picture of the health of the environment.