To diagnose application performance bottlenecks, it is essential to collect metrics from the core components of the IT infrastructure that directly impact application responsiveness. Network latency measures the delay in data transmission between components (e.g., application server and database), which is a primary factor in slow transaction times. Storage queue depth indicates the number of I/O requests waiting for the storage system to process. A high queue depth signifies that the storage is overwhelmed and cannot keep up with the application's demand, creating a significant I/O bottleneck.

C. Number of business units: This is an organizational or business metric, not a technical performance indicator used for diagnosing infrastructure bottlenecks.

D. Office internet bandwidth quota: This is a policy-based limit on data usage, not a direct measure of real-time performance. While available bandwidth is a relevant metric, a quota is a ceiling.

1. Storage Queue Depth (B): HPE. (2017). HPE 3PAR StoreServ Storage Best Practices for VMware vSphere 6.5. Page 31, Section "Performance Monitoring". The document explains that monitoring the device queue depth on an ESXi host is critical for identifying storage performance contention, a common cause of application slowdowns. It states, "If the device queue depth (DQLEN) is consistently high, it indicates that the storage array is not able to service the I/O requests from the host in a timely manner."

2. Network Latency (A): VMware, Inc. (2022). Performance Best Practices for VMware vSphere 8.0. Page 71, Chapter 4, "Network Performance". This official guide, widely used for applications running on HPE hardware, emphasizes the importance of latency: "Network latency is a very significant component of network performance. In some cases, a small number of CPU cycles can be traded for a significant reduction in network latency." This highlights its direct impact on application speed.