Application-level metrics enable IT teams or developers to investigate and diagnose application issues. Depending on what we know about our application, we can set up some critical metrics for monitoring. A few important application-level metrics are listed here:

• Average application response time: This refers to the performance of our application. In other words, it is the amount of time taken by website to return a HTTP/HTML/request to an end user. These metrics will give us a sense of the performance from the end user's side. There are many tools by which we can monitor the application response time. SolarWinds is an example of a tool that can identify the root-cause of response time issues and improve slow server and application performance:

The following screenshot gives an example of the top ten application response times:

• Peak response time: The peak response time metric enables IT teams to identify slow elements within the application. It help to improve the overall application performance. For example, one particular element might take 10 seconds to load, while the average response time is much less than that.
• Error rate: This metric enables companies and IT teams to understand when the application will fail. It is expressed as a rate of error per unit. The error rate is a calculation of how many requests failed compared to the total requests.
• Concurrent users: This refers to the number of people using the resources within a predefined period of time. Each web server has a limit on the maximum number of concurrent users. It may correlate with the average response time because if the concurrent users increase on a web server, this will increase the average response time as well. Moreover, with an increase in concurrent users, a system will need more resources, such as bandwidth, memory or CPU cycles, depending on the hardware of the server and the type of requests that are being served. 
• Requests per second (RPS): This indicates how many requests (such as web pages, images, video, audio files, or resources from databases) are being sent to the backend server every second. The average RPS rate may differ from company to company depending on how much the high load is per application. For example, content such as images or HTML (static content) is served by a web server such as Apache or NGINX. Content that needs to query the database, however, will require more resources, as it needs to connect to the web server and the backend database as well.
• Throughput: This is the amount of data, information, requests, and packets that pass through a system. It is a measure of how much bandwidth is required to handle concurrent users and requests. A higher throughput value is a good thing for an application, as it indicates that an application can handle an increasing number of concurrent users.