Help Center> Elastic Load Balance> FAQs> Load Balancing Performance> How Do I Determine the Server Response Time Based on Monitoring Data and Logs?

Updated on 2023-05-18 GMT+08:00

View PDF

How Do I Determine the Server Response Time Based on Monitoring Data and Logs?

For HTTP and HTTPS load balancing, you can view the average server response time through monitoring metric and view the response time of each request from access logs.

On the ELB console, click the name of the load balancer.
On the Monitoring tab page, select an HTTP or HTTPS listener added to the load balancer.

Check the Average Server Response Time metric to view the average time that backend servers respond to requests routed by the load balancer.

**Table 1** Average response time
Metric	Definition
Average Server Response Time	Average time that backend servers respond to requests from the load balancer (This metric is available only when the frontend protocol is HTTP or HTTPS.) The process starts when the load balancer routes the requests to backend servers and ends when it receives responses from backend servers. Unit: ms

Check access logs to view the response time of each request.

The request_time, upstream_connect_time, upstream_header_time, or upstream_response_time fields in the access log reflect the time required for a load balancer to route a request to the corresponding backend server.

**Table 2** Parameter description
Field	Description
request_time	Request processing time in seconds, that is, the duration from the time when the load balancer receives the first request packet from the client to the time when the load balancer sends the response packet
upstream_connect_time	Time taken to establish a connection with the server, in seconds with a milliseconds resolution When the load balancer attempts to retry a request, there will be multiple connection times. If the request is not correctly routed to the backend server, a hyphen (-) is displayed as a null value for this field.
upstream_header_time	Time taken to receive the response header from the server, in seconds with a milliseconds resolution When the load balancer attempts to retry a request, there will be multiple response times. If the request is not correctly routed to the backend server, a hyphen (-) is displayed as a null value for this field.
upstream_response_time	Time taken to receive the response from the server, in seconds with a milliseconds resolution When the load balancer attempts to retry a request, there will be multiple response times. If the request is not correctly routed to the backend server, a hyphen (-) is displayed as a null value for this field.