Interpret HTTP server test results

Use ThousandEyes tests to analyze network conditions during a ThousandEyes HTTP Server test. Review network metrics, compare results with the recent baseline, and determine whether a performance issue is related to the network path or the target application.

View HTTP server test results and metrics

  1. In Splunk Observability Cloud, select Digital Experience and then Synthetic tests with network data.

  2. Select the HTTP server test that you want to investigate.

  3. Review the run trends and network metrics from the Network insights tab.

Interpret HTTP server test run results

The following table lists the metrics for HTTP server tests:
Metric Description
Loss The percentage of probe packets dropped between the ThousandEyes test location and the target server. Higher packet loss can contribute to slow page loads, failed requests, and failed API calls.
Latency The round-trip time for data to travel between the test location and the target server. High latency indicates that more of the application's response time is being spent waiting on the network.
Jitter The variation in latency measurements. High jitter indicates an inconsistent network path.

Use the Network insights tab together with the test's application-layer results to narrow down the cause of an issue.

  • If response time is high and network metrics are also elevated, investigate the network path, test location, and destination.

  • If response time is high but network metrics are near the recent baseline, investigate the target application or server.

  • If packet loss, latency, or jitter is elevated for one location, investigate that location's path to the target.

  • If the issue appears across multiple locations, investigate the target service, shared network dependencies, or a regional network problem.

Compare results with recent baseline

Each network metric is compared with the P75 for the previous seven days. The P75 represents the value below which 75% of the metric measurements occurred during the rolling seven-day period.

For example:

  • Seven-day P75 packet loss: 3%

  • Current run packet loss: 4.5%

In this example, the current run has higher packet loss than the recent baseline and might require investigation. A metric that is higher than the baseline does not necessarily mean that the test failed. It indicates that the result was unusually high compared with recent runs.

HTTP server test metrics

When you create or import ThousandEyes tests, Splunk Observability Cloud ingests the following OpenTelemetry network metrics from ThousandEyes tests:
Metric Layer Description

http.server.request.availability

HTTP

0% on error, 100% otherwise

http.client.request.duration

HTTP

Time to First Byte (seconds)

http.server.throughput

HTTP

Download throughput (bytes/sec)
http.server.throughput HTTP DNS Time/Resolution (ms)

network.latency

Network

Max round-trip latency (seconds)

network.loss

Network

Packet loss (%)

network.jitter

Network

Latency variation (ms)